Open access peer-reviewed chapter - ONLINE FIRST

Wearable Sensors and Devices: Enhancing Human Life

Written By

Chris Cheng Zhang, Mark Yining Liu, Sky Yangziming Han, Amirreza Sedaghat, Kevin Zhang and Haoyang Wang

Reviewed: 05 May 2025 Published: 14 June 2025

DOI: 10.5772/intechopen.1010293

Current Developments in Biosensors and Emerging Smart Technologies IntechOpen
Current Developments in Biosensors and Emerging Smart Technologie... Edited by Selcan Karakuş

From the Edited Volume

Current Developments in Biosensors and Emerging Smart Technologies [Working Title]

Selcan Karakuş

Chapter metrics overview

27 Chapter Downloads

View Full Metrics

Abstract

Wearable sensors are changing almost every sector including healthcare, fitness, industrial, safety and even entertainment. These devices are intended to be comfortably affixed to the body and to continually monitor physiological, environmental and motion-related signals. This chapter first introduces the definition of wearable sensors/devices, including various hardware architectures and development platforms, and the differences from conventional sensors. There are currently thousands of different sensors available. For example, widely used gyroscope sensors, acceleration sensors, light sensors, temperature sensors, etc. These applications have been used in industrial, medical, aerospace and other fields. Data collection methods and machine learning (ML) techniques using wearable sensors today will also be described. The importance of low-power design, efficient data handling, integrating artificial intelligence into wearable sensors and the relationship to the Internet of Things (IoT) is explored. Finally, this chapter discusses the challenges and future of this quickly developing technology. At the end, we answer the question: will wearable sensors disappear in the future or will they be used more?

Keywords

  • wearable sensor
  • hardware architectures
  • data collection
  • machine learning
  • fall prevention

1. Introduction

Today, there are thousands of different sensors used in many fields. For example, distance sensors or ultrasonic sensors are installed in vehicles to avoid collisions. Fog sensors and heat sensors can be installed indoors to detect fires and trigger fire alarms. Light sensors and cameras (which can also be a type of sensor) can be used in cleaning robots. Carbon monoxide (CO) sensors are used to detect gas leaks. Fuel level sensors are installed in gas tanks to detect the oil level (Figure 1).

Figure 1.

Left: three-axis gyroscope + accelerometer; center: carbon monoxide (CO) sensor; right: ultrasonic sensor.

However, for wearable sensors, although they share lots of features similar to traditional sensors, but they do have their own special properties. Wearable sensors are devices specifically designed to monitor physiological, environmental or motion-related parameters while being comfortably worn on the body. These sensors have transformed how data is collected and analyzed, offering real-time, continuous and non-invasive data without the necessity for human monitoring or intervention. Their key attributes are portability, energy efficiency and adaptability, making them versatile tools applicable across numerous fields, from everyday personal use to industrial and medical applications (Figure 2).

Figure 2.

Left: smartwatch, right: fitness tracker.

Unlike traditional sensors, wearable sensors are customized for integration into daily life. They are often embedded in accessories such as smartwatches, fitness trackers or clothing. For example, a fitness tracker can monitor step counts, heart rate and sleep patterns, while a body-worn movement sensor can record body motion in real time for virtual representations. These devices have become indispensable, enhancing convenience and functionality in previously unattainable ways.

1.1 Comparison between traditional sensors and wearable sensors

Wearable sensors and traditional sensors share the common goal of data collection but differ significantly in their design, functionality and applications. Traditional sensors are typically static and installed in fixed locations or embedded within machines. They are oriented for specific tasks, such as detecting obstacles with ultrasonic sensors in vehicles or monitoring gas leaks with carbon monoxide sensors at home. These sensors are effective in controlled environments where they do not require mobility or adaptability. By contrast, wearable sensors are portable and specifically designed for continuous use on the human body. For instance, a smartwatch with an accelerometer and gyroscope can track physical movements in real time, enabling users to monitor their activity seamlessly throughout the day.

Another key difference lies in the applications of these sensors. Traditional sensors are generally task-specific and in fixed settings, like gas detectors at home. On the other hand, wearable sensors are multifunctional and user-centric. They can adapt to dynamic environments and monitor a wide range of variables, such as heart rate, blood oxygen level and activity. This versatility makes wearable sensors invaluable in applications where traditional sensors would be limited by their static usages.

The interaction between users and these devices also varies. Traditional sensors often require manual operation and human monitoring, with data retrieval necessitating specialized tools. However, wearable sensors are designed to integrate seamlessly into daily routines, allowing users to review the data collected easily. Their user-friendly interfaces and wireless connectivity enable individuals to access real-time data effortlessly through the device itself or smartphones, enhancing convenience and usability.

Design features further distinguish wearable sensors from traditional sensors. Traditional sensors prioritize durability and precision for specific tasks but are not optimized for mobility or comfort. On the other hand, wearable sensors emphasize compactness, lightweight and long-lasting battery life. These characteristics ensure that users can wear them comfortably for extended periods while maintaining accurate, real-time monitoring. Table 1 summarizes the key differences between wearable and traditional sensors.

FeatureTraditional sensorsWearable sensors
MobilityStatic, fixed locationsPortable, worn on the body
applicationTask-specific, controlled environmentsMultifunctional, dynamic environments, user-centric
Typical use casesObstacle detection (ultrasonic sensors in vehicles), gas leak monitoring (CO sensors)Activity tracking (accelerometer, gyroscope), health monitoring (heart rate, blood oxygen)
User interactionOften manual operation, specialized data retrieval toolsSeamless integration into daily routines, user-friendly interfaces, wireless connectivity (smartphone access)
Design prioritiesDurability, precision for specific tasksCompactness, lightweight, long battery life, comfort
Data accessMay require direct connection or manual readingReal-time data access via device or connected apps

Table 1.

Comparison of wearable and traditional sensors.

Advertisement

2. Hardware architectures of wearable sensors

2.1 Introduction: The compact and constrained

Today, wearable sensors are complex embedded systems designed for continuous, on-body monitoring, constrained by power and size limitations. These sensor devices are characterized by specific limitations that greatly influence their built-in hardware design. In contrast to computers or smartphones wearable sensors function within the confines of performance capabilities they need for remarkable energy efficiency and strict limitations regarding their physical size and shape. These limitations do not exist in isolation; instead, they are close. Require evaluation and creative design approaches for all hardware elements.

Wearable sensors have been getting better over time. They are purposely kept limited in terms of performance to save energy and stay compact in size. They are designed with duties in mind, like monitoring body signals or collecting information. This focused approach differs from the computing capabilities of smartphones or laptops. Battery life is a priority when it comes to devices, as it requires careful consideration of ultra-low power design principles at all stages of the hardware architecture process. From choosing components to executing sophisticated algorithms with energy efficiency in mind. Also important is the design of sensors to ensure user comfort and seamless integration into life while maintaining appealing esthetics. When creating sensors, the size and weight matter a lot. Determine the need for very small components and new packaging methods as well as flexible substrates, in certain situations. This section will explore the hardware elements that shape the structure of sensors, which include designing processing units, memory management, communication module implementation, sensor interface complexities and essential power management techniques.

2.2 Body signal acquisition and transmission principles

Wearable devices capture a diverse range of physiological signals using specialized sensors, which are then typically processed and transmitted wirelessly. Understanding the basic principles behind these processes is crucial for appreciating the capabilities and limitations of wearable technology.

2.2.1 Common sensing modalities

Different sensors operate on distinct physical or chemical principles to measure specific bodily functions:

  • Photoplethysmography (PPG): Commonly used for heart rate and blood oxygen (SpO2) monitoring. This optical technique involves emitting light (usually LED) into the skin and measuring the amount of light reflected or transmitted back to a photodetector. Changes in blood volume within the microvascular bed due to the cardiac cycle modulate the light absorption/reflection, allowing estimation of heart rate. Differential absorption at specific wavelengths (e.g., red and infrared) enables SpO2 calculation [1].

  • Electrocardiography (ECG): Measures the electrical activity of the heart via electrodes placed on the skin. These electrodes detect the potential differences generated by the depolarization and repolarization of heart muscle cells during each heartbeat, providing detailed rhythm information useful for detecting arrhythmias [2, 3].

  • Electromyography (EMG): Detects the electrical potential generated by muscle cells when they contract. Similar to ECG, electrodes placed on the skin over the muscle measure these signals, which can be used to assess muscle activity, fatigue or control prosthetic devices [2, 3].

  • Inertial Measurement Units (IMUs): Typically combine accelerometers and gyroscopes. Accelerometers measure linear acceleration (including gravity), often using the displacement of a micro-machined proof mass. Gyroscopes measure angular velocity, often using the Coriolis effect on a vibrating structure. Together, they provide detailed motion and orientation data used for activity tracking, gait analysis and fall detection [3].

  • Electrochemical sensors: Used for detecting specific chemical analytes. For instance, continuous glucose monitors (CGMs) often use an enzyme (like glucose oxidase) on an electrode. The enzyme reacts with glucose in the interstitial fluid, producing an electrical signal proportional to the glucose concentration [4].

These raw analog signals are typically amplified, filtered to remove noise and then converted to digital signals by analog-to-digital converters (ADCs) within the wearable device for processing by the microcontroller [5].

2.2.2 Data transmission

Once acquired and processed locally to some extent, data is often transmitted wirelessly to a companion device (like a smartphone) or directly to the cloud. Common protocols include:

  • Bluetooth Low Energy (BLE): Optimized for low-power consumption over short ranges (up to 100 m). Ideal for periodic data transfer from sensors to smartphones in applications like fitness tracking and basic health monitoring [6, 7].

  • Wi-Fi: Offers significantly higher data rates and standard internet connectivity but consumes more power. Suitable for wearables requiring larger data transfers or direct cloud connection when near an access point [8].

  • Cellular (e.g., Long-Term Evolution (LTE) and 5G): Provides wide-area connectivity independent of a smartphone, enabling standalone functionality for devices like advanced smartwatches or remote patient monitoring systems, albeit with higher power demands [8, 9].

The choice of transmission technology depends on the specific application’s requirements for data volume, frequency, range and battery life constraints.

2.3 Low-power architectures: The Foundation of wearable design

The main idea driving the creation of sensor devices is the crucial need for low energy usage rather than just being a nice feature to have; it is essential for extending battery life span and ensuring comfort during wear while enabling continuous monitoring across various use cases like healthcare and fitness tracking tasks alike. Bringing about an energy-efficient operation in wearables demands a thorough strategy that considers all levels of the device’s design, from picking specific transistors to laying out the overall system structure carefully.

One key method to lower power usage is to reduce the power consumption of the circuit from the battery. A significant portion of power consumption comes from the switching of transistors within circuits. Techniques like clock gating, which disables the clock signal to inactive circuit blocks and operand isolation, which prevents unnecessary switching in functional units by isolating input data until it is needed, are some of the strategies used for reducing this switching power [10]. Additionally, the development of efficient algorithms is important as it results in switch operations and less energy consumption. Leakage power or static power occurs when there is a flow of current in circuits that are not actively switching functions. To reduce this type of power consumption and save energy, it is crucial to choose transistors with leakage currents and implement power gating technology for unused circuit blocks to completely cut off the power supply when not in use. Furthermore, voltage scaling is another method to decrease power consumption as power is directly proportional to the square of the supply voltage, meaning that even small decreases in voltage can result in significant energy savings [11]. Dynamic voltage and frequency scaling (DVFS), used for managing power consumption efficiently, adjusts the voltage and frequency of the processor in time according to the workload. This enables the system to function at reduced power levels during times of low-performance demand.

The development of energy devices frequently requires strong teamwork between hardware and software experts, known as hardware-software co-design. Factors like the usage pattern of the application are used to determine the most suitable hardware components to enhance power efficiency overall. For instance, a program that occasionally collects information can take advantage of hardware features that enable deep sleep modes with low power consumption when not in use. In order to assess and compare hardware platforms for low power wearable uses in fields such as biomedicine effectively and efficiently, benchmark suites like BiomedBench have been created [12]. BiomedBench offers a collection of end-to-end TinyML (Tiny Machine Learning) biomedical applications that cover idle timeframes in addition to acquisition and processing stages. This standardized set enables an evaluation of energy efficiency across cutting-edge low-power platforms. This standardization aids hardware developers in recognizing design elements that affect performance and assists application designers in choosing the appropriate deployment platform [13].

Additionally, studies have grouped the methods of conserving power in medical gadgets into a classification that covers organizing tasks, compressing signals, managing clocks and power. Scheduling techniques for tasks such as optimizing duty cycles and evenly distributing workloads help decrease the device’s power usage. Signal compression methods such as Compressive Sensing and Joint Compressive Sensing reduce the data that requires processing and transmission, resulting in energy savings. Some other strategies that are used for managing the clock in devices help in controlling power usage through methods like adjusting frequency and gating the clock signal. In addition, power management tactics involve being aware of the power status of the devices and shutting off the circuits to regulate energy consumption [14]. Choosing and implementing these low-power techniques is essential for ensuring the energy efficiency needed for sensor applications.

2.4 The Core components of wearable technology

The overall architecture of a wearable sensor system integrates several key components, each playing a vital role in data acquisition, processing and communication. Figure 3 presents a block diagram showing the typical arrangement of these components.

Figure 3.

Block diagram of a typical wearable sensor system. Key components include sensors (e.g., accelerometer and gyroscope), a microcontroller (containing the processor core, memory and peripherals), a power management unit, communication modules (e.g., Bluetooth and Wi-Fi) and user interface elements (e.g., display and buttons).

The core processor in a wearable sensor acts like a “brain” that performs several key functions: it processes sensor data, executes algorithms for analysis and controls communication and user interface elements. Selecting a power-efficient processor is essential for maximizing battery life in these devices.

The ARM Cortex-M series of microcontrollers has become an option, for devices due to their energy-efficient design and suitability, for low-power wearable applications [15]. Among these processors is the Cortex-M0+, known for its power efficiency and compact size, which makes it perfect for battery-operated wearables [15]. Despite being compact and energy-efficient, the Cortex-M0+ has the ability to handle 32-bit processing tasks effectively. This makes it suitable for sensor applications while still maintaining power efficiency similar to that of 8 or 16-bit processors. The trade-off between performance and power consumption makes the M0+ suitable because it provides sufficient processing power for many wearable sensor tasks without the higher power drain of more powerful processors like the M4 or M7. Its functionalities are tailored for sensors and wearable gadgets, reinforcing battery longevity and cost savings. Although the Cortex-M series includes processors, like M4 and M7, that are better suited for use than the Cortex-M0+, they come at the expense of power consumption. This makes them more ideal for powering devices that require increased performance [15].

In addition to the ARM Cortex-M series of processors used in sensors mentioned by [16] study, there are other types of processing units employed based on the specific needs of the application field. Microcontroller units (MCUs), which are processors providing a mix of performance and energy efficiency, are commonly utilized for an array of functions in wearables. Digital signal processing (DSP) chips are made for handling signal processing tasks commonly found in sensors, like cleaning up sensor signals and pulling out data from raw information. On the other hand, application-specific integrated circuit (ASIC) chips are custom-designed to match the functions of wearable gadgets [16]. Although ASICs can provide benefits in terms of energy efficiency and speed, for their purposes. However, they usually come with high development expenses and limited adaptability when compared to off-the-shelf, more general-purpose processors. As stated more clearly, the disadvantages of ASICs are their high initial development cost and their inflexibility to be reprogrammed for different tasks. For example, a designed ASIC meant to accelerate data-driven classification in a wearable device showcased power usage that was significantly lower by several degrees than a software version running on a standard low-power processor.

The latest wearable gadgets, like smartwatches, are now incorporating 64-bit ARM processors to enhance their capabilities and provide users with richer features and more complex interactions [17]. 64-bit processors offer advantages over their 32-bit counterparts, including the ability to address more memory (crucial for complex applications), handle larger data sets more efficiently and execute more complex instructions. While these processors bring performance improvements to the devices, allowing for advanced functions and better user experiences, they also pose the challenge of efficiently handling power usage and heat dissipation [17]. Research studies indicate that ARM processors in watches go through design iterations using assertive scaling methods to enhance performance. However, the optimal balance between performance improvements and managing power consumption and heat remains a concern. When comparing 64-bit ARM processors in watches, it is evident that no single processor dominates all areas, but instead each shines in specific features. This showcases the trade-offs involved in designing processors for technology.

Table 2 provides a comparison of different microcontroller options commonly used in wearable devices, highlighting their trade-offs in terms of power consumption, processing capabilities, cost and typical applications.

MicrocontrollerPower consumptionProcessing powerCostTypical applications
ARM Cortex-M0+Very Low (10–100 μA/MHz)Low-Medium (20–50 MHz)Low ($0.50–$2)Basic fitness trackers, simple sensor hubs and low-power IoT devices
ARM Cortex-M4Low-Medium (50–200 μA/MHz)Medium (80–200 MHz)Medium ($1–$5)Smartwatches, more complex fitness trackers, industrial sensors and basic audio processing
ARM Cortex-M7Medium-High (100–300 μA/MHz)High (200–600 MHz)Medium-High ($3–$10)Advanced smartwatches, medical devices with signal processing, complex sensor fusion
DSPsVariable (depends on architecture)High (specialized for signal processing)VariableAudio processing, noise cancelation and real-time sensor data filtering
ASICsVery Low (highly optimized)High (for a specific task)High (initial NRE cost), Low (per unit at high volume)Highly specialized applications requiring extreme power efficiency or performance (e.g., custom health monitoring chips)

Table 2.

Comparison of Microcontroller Options for Wearable Devices.

2.5 Managing data in limited resources environments

Efficient memory management is absolutely critical for the operation of compact, battery-powered gadgets, directly impacting their performance and longevity [18]. Effectively handling these resources is vital for ensuring the seamless operation and effectiveness of sensors. These devices utilize memory technologies tailored to specific needs and functionalities, each with inherent trade-offs.

SRAM (Static RAM), widely recognized for its speed and easy accessibility compared to other memory types, is also volatile and power-hungry. It only maintains data while powered on—often found in wearables for short-term data storage, like temporarily storing sensor data before processing and serving as cache memory to hasten retrieval of commonly used information. Flash memory, in contrast, is a non-volatile memory technology, capable of storing data without a constant power supply. It is often utilized in gadgets to hold the device’s software code and larger sets of data, like recorded sensor information. Two primary categories of flash memory exist: NOR flash, which enables fast random access and is commonly employed for executing code and NAND flash, which provides greater storage capacities at a lower cost per bit, making it preferred for storing larger datasets. Ferroelectric RAM (FRAM) offers another alternative. It combines the non-volatility of flash memory with the speed, energy-efficient operation and strong durability (ability to endure numerous writing cycles without performance degradation) of SRAM [19]. However, FRAM (Ferroelectric RAM) may have lower capacity and a potentially higher cost than flash. FRAM is especially suitable for devices that need to frequently record data without using up too much energy.

To efficiently handle the memory and storage capacities in these devices, various tactics are utilized. Data compression methods are crucial for cutting down the volume of data requiring storage, consequently decreasing memory consumption and power usage (as there is less data to write and read) [14]. Selecting optimal data structures and algorithms can also aid in lowering the memory requirements of applications and enhancing data retrieval speeds. Direct memory access (DMA) is a hardware capability that enables components, like sensors and communication modules, to move data directly to or from memory without needing involvement from the central processing unit (CPU) [18]. By offloading data transfer duties from the CPU, DMA significantly reduces the CPU’s burden and overall energy usage. This permits the CPU to conserve power by entering low-power sleep states or focusing on other processing activities [20]. Studies have looked into creating compact storage solutions for gadgets that utilize DMA and the serial peripheral interface (SPI) protocol to enable ongoing monitoring [20]. These solutions frequently arrange data into time-based segments or files to minimize control overhead and simplify the handling of scheduled data captures [20]. Research, such as that found in ScholarWorks, has highlighted ultra-low-power NAND flash memory as an excellent storage choice for sensor networks due to its notably reduced energy usage per byte when contrasted with traditional serial flash memory choices [21]. In essence, the development of memory and storage systems for gadgets necessitates a careful balancing of factors such as memory capacity, power usage, cost and the unique demands of the target applications.

2.6 Efficient energy usage in continuous sensing to stay active without draining battery power

In the field of sensor applications, like health and wellness monitoring devices, as mentioned in a study by FutureTrends research [7], it is essential to have sensors that can continuously track environmental data over time to ensure accurate monitoring results without draining the battery life of the wearable devices too quickly due to constant data processing demands. Time-consuming processing tasks are crucial for preserving battery life and enabling sensing operations.

A key strategy for saving energy is duty cycling [14]. This approach includes turning the sensor and processing unit on and off at intervals to collect and analyze data efficiently while conserving power when not in use. The duty cycle refers to the proportion of time the devices are active compared to the total duration. It can be customized according to the needs of a particular application and how frequently the parameters being monitored are likely to fluctuate. In situations where there are minimal changes in the data being analyzed, one can use a low-duty cycle to cut down on power usage significantly. Another important method to save energy is by utilizing signal processing algorithms designed for low-power consumption. With the help of algorithms that are less computationally intensive, the processing unit can carry out data analysis tasks with fewer operations, resulting in lower power consumption. For instance, optimized filtering techniques can be employed to eliminate noise and efficient methods for feature extraction can reduce the volume of data that needs processing. Utilizing hardware acceleration is crucial for enhancing energy efficiency in computing systems dedicated to sensing [16]. Specialized hardware components, or accelerators, are tailored to execute particular demanding functions, like signal processing or machine learning inference, with lower power consumption compared to general-purpose processors at comparable performance levels. By delegating these tasks to dedicated hardware units, the primary processing unit can operate at a reduced power level for extended durations.

In event-driven sensing methodology, data acquisition and processing are initiated upon detection of a significant change or event in the environment being monitored [22]. This strategy prevents unnecessary sampling and processing during periods of stability in the variables, resultantly leading to considerable energy conservation benefits. In a scenario, a wearable accelerometer might engage in analysis only when it detects a sudden motion change, suggesting a fall occurrence. When dealing with sensors in different situations in a network, data aggregation is a useful method to save power [22]. By handling and combining information from multiple sensors within the network before sending it wirelessly, the overall data transmission volume can be notably decreased, resulting in energy conservation for both processing and communication tasks. Studies have also delved into the idea of using energy for context recognition and forecasting on smartwatches [23]. Through the use of abstraction to transform raw sensor data into valuable information entities and carrying out predictive functions directly on the device itself, one can reduce the necessity for frequent data transmission to external devices. This leads to decreased power usage [23]. Additionally, energy harvesting technologies, such as solar panels, thermal converters and kinetic energy harvesters, have the capability to complement battery power in gadgets, allowing for extended or potentially continuous functionality [11]. By harnessing energy from the surrounding environment or the user’s own body to generate electricity, these advancements can lessen dependence on batteries and prolong the functionality of wearable sensors.

2.7 Connecting sensors and gathering data; bridging the gap between the physical and digital realms

Wearable sensors can connect with the environment, transforming real-life events into digital data through their sensor interfaces and data collection methods, which play a crucial role in their efficiency. Various communication protocols are frequently employed to link sensors with the processing unit in gadgets [24]. Among these are I2C (Inter-Integrated Circuit) and SPI.

The I2C protocol is commonly used for linking lower-speed devices to a microcontroller. It involves two lines: a data line (SDA – Serial Data) and a serial clock line (SCL – Serial Clock). One of the benefits of I2C is its simplicity, as it only requires two wires for communication. Additionally, it can handle multiple slave devices on one bus, with a master device overseeing the communication process. In research and when developing systems for prototyping purposes, modularity and solid backing in integrated circuits are frequently favored [24]. Nevertheless, SPI generally provides faster data transfer speeds than I2C but may present address restrictions that come into play when connecting with numerous sensors [24].

SPI is a communication protocol that operates using four wires: a serial clock (SCK), a master out slave in (MOSI), a master in slave out (MISO) and a slave select (SS). Compared to I2C (inter-integrated circuit), SPI provides faster transfer speeds and enables full-duplex communication, where information can be transmitted and received concurrently [25]. However, it needs more pins than I2C. It also does not include a built-in addressing system for multiple slave devices, usually needing a separate slave select line for each sensor [26]. The decision between I2C and SPI relies on the needs of the application, taking into account aspects like the quantity of sensors, the desired data speeds and the available microcontroller pin count (Table 3).

FeatureI2CSPI
Speed (Data Rate)Lower (up to 3.4 Mbps)Higher (several Mbps or MHz)
Complexity (Number of Wires)Simpler (2 wires)More complex (4+ wires)
Complexity (Protocol)More complex (addressing, arbitration)Simpler (no arbitration)
Power consumptionGenerally lowerGenerally higher
Scalability (Number of Devices)High (multiple slaves on one bus)Limited by SS pins
Noise immunityLower (shared bus)Higher (separate lines)
Full/Half-duplexHalf-duplexFull-duplex

Table 3.

Comparison of I2C and SPI Protocols.

Analog-to-digital converters (ADCs) are essential components in sensor technology, as they convert analog output signals into digital values that can be read by a microcontroller [16]. When it comes to sensors, important ADC specifications to consider are resolution (the bit depth for representing analog values), sampling rate (measurement frequency) and power usage efficiency. The resolution influences the quality of the representation of the analog signal and the sampling rate controls how often sensor data is captured. In devices like wearables, due to power limitations, it is important to pick ADCs that provide the required performance features with minimal power consumption. Usually, before sensor signals are sent to an ADC, they go through signal conditioning circuitry [5]. The circuitry might have amplifiers to boost the signal strength and filters to minimize any disturbances or interference present in the signal from the sensors before converting it into a digital format for dependable digitization by the ADC.

Wearable sensors can wirelessly connect with devices like smartphones, tablets or cloud servers. This feature allows data to be sent for storage, analysis and user engagement purposes. Wearable devices use communication technologies that have unique qualities suited for different applications [8].

Bluetooth low energy (BLE) is a technology in wearable devices known for its exceptional energy efficiency due to the minimal power it consumes [7]. It is frequently utilized for close-range communication with smartphones and neighboring gadgets to enable data exchange for fitness trackers and health monitors, such as smartwatches. Wi-Fi provides higher data speeds and has a greater coverage distance than BLE but generally requires more power (Wireless Communications). The devices that are wearable and require high bandwidth, for instance, Internet-connected smartwatches, use them. Near-field communication (NFC) is a low-power communication technology with a short range and is used in many applications for purposes such as payments and data exchange between nearby devices, as suggested in [8]. Wearable devices that need to stay connected across wide areas and function independently, without depending on a smartphone, often use cellular technologies like long-term evolution (LTE) and the emerging 5G networks, according to Wireless Communications research [8]. Although these cellular technologies provide a wider coverage range compared to short-range options, they do consume more power as a trade-off feature. Moreover, this technology, having appeared, makes it possible to support a large number of wearable sensors and devices in industrial and IoT environments, as found in the wireless communications research (Table 4) [27].

TechnologyRangeData ratePower useApplications
BLEUp to 100 mUp to 2 MbpsVery LowFitness trackers, health monitors, smartwatches (data sync)
Wi-FiUp to 100 m (indoor)Up to several hundred MbpsHighSmartwatches (internet), some medical devices
NFCFew cmUp to 424 kbpsVery LowContactless payments, data transfer, access control
Cellular (LTE)Wide area (km)Up to 100+ Mbps (downlink)HighSmartwatches (cellular), remote health monitoring
Cellular (5G)Wide area (km)Up to GbpsHigh (potential for IoT opt.)future connected wearables

Table 4.

Comparison of wireless communication technologies for wearables.

The design of the network linking sensors can differ based on the usage scenario employed. In a point-to-point setup, the wearable gadget communicates directly with a device, like a smartphone, using Bluetooth technology. In a star network configuration, multiple wearable gadgets link to a hub, which could be a base station or smartphone, serving as a gateway. Mesh networks enable devices to not only communicate with a central point but also exchange data among themselves while traversing the network. In situations where there are sensors connected to each other, using devices such as smartphones as intermediaries can enhance reliability and scope. These smartphones act as bridges between the sensors, receiving data through short-range methods like Bluetooth and then transmitting it to cloud servers or centers through networks such as Wi-Fi or cellular connections [9].

2.8 Power management: Extending the lifeline of wearable devices

The longevity of sensors heavily relies upon the battery life they possess; hence, the implementation of power management strategies becomes crucial to prolong their usage time effectively by minimizing energy usage in all operational facets of the device.

One of the methods employed in managing power is dynamic voltage and frequency scaling (DVFS), which regulates the voltage and clock frequency of the processor based on the workload. In the case of the device operating under conditions that require more power, the voltage and frequency are increased to provide the necessary performance [10]. However, in times of low activity or when the device is idle, the voltage and frequency can be lowered to conserve power. Another important method is power gating, which entails turning off the power supply to sections of circuits not in use. By doing this, static leakage power can be greatly minimized, which plays a significant role in the overall power usage of current integrated circuits. Power management integrated circuits (PMICs) are specifically designed integrated circuits that effectively allocate power in wearable devices [28]. In their role, they monitor the charging of the battery and control the voltage needed for different components as well as include different power save modes for greater energy efficiency.

In the case of wearable devices, energy harvesting technologies can potentially supplement or even replace batteries, according to [11]. These technologies get energy from the environment or the human body, for example, small photovoltaic cells that collect energy from solar power, thermoelectric generators that work on the principle of converting heat energy from body warmth or kinetic energy from body movements through piezoelectric or electromagnetic devices. Although the available power is usually small, it may be sufficient for operating low-power sensors or extending the battery life of some gadgets. Recent research has also been conducted on the improvement of the power management policies to enhance the energy efficiency in sensor networks. For example, HEEPS (Hybrid Energy-Efficient Power manager Scheduling) integrates dynamic power management policies and DVFS to enhance power consumption [10]. These latest strategies are aimed at predicting and analyzing the sensor node behaviors in order to use the energy conservation techniques properly.

2.9 Hardware design and the rise of artificial intelligence (AI) and machine learning in wearables

In the realm of sensor technology, AI and ML are being integrated into devices at a rise [29]. The reason for this shift is to enable the wearables to work autonomously by analyzing the sensor data in real time, providing personalized insights and responding quickly to the user’s needs or variations in the environment. Hardware architectures are being tailored and refined to meet the increasing computational needs for on-device AI processing. This adaptation is necessary because machine learning algorithms, especially complex ones used for tasks like real-time activity recognition or physiological anomaly detection, often involve intensive computations (such as matrix operations or iterative optimizations) that demand efficient hardware acceleration to run within the strict power budgets of wearable devices [29, 30].

A significant improvement is the integration of new hardware accelerators, such as neural processing units (NPUs) and tensor processing units (TPUs), which are designed to excel at matrix multiplications and other computations that are essential for machine learning algorithms [16]. When run on general-purpose CPUs, these accelerators may enhance the performance and energy efficiency of AI tasks considerably. Additionally, there are efforts to optimize memory architectures for efficiently supplying data to these processing units during machine learning calculations, thus reducing bottlenecks and enhancing overall efficiency. Given the power usage limitations of wearable gadgets, there is an emphasis on crafting and picking energy-efficient AI algorithms, commonly known as TinyML [13]. These algorithms are developed to use them in wearable devices with limited resources, and for this reason, these algorithms are small in size and less computationally complex. These algorithms are developed to occupy less memory space and need fewer computations.

The integration of AI and ML in wearable devices also has numerous benefits. On-chip AI enables the evaluation of sensor information without the help of other servers. This leads to decreased delays in the flow of information and provides instant output to the users. Furthermore, decreasing the data processing burden at the device level also improves the privacy of the users. Reducing the dependency on the cloud can increase the robustness and dependability of applications, particularly in areas with poor network coverage. However, the integration of AI and ML in wearables is not without some challenges. These are the challenges, though: limited resources and memory of these devices and the necessity of designing power-conservative devices. Despite all these, AI and ML are already common in wearables for functions such as identifying movements (for instance, identifying different forms of exercise), checking health (for instance, for falls or irregular heartbeats) and assisting people with information and suggestions.

2.10 Hardware design summary

Wearable sensor hardware designs are a combination of computer engineering and material science, with human-computer interaction mixed in. To create them, you have to find that sweet spot between performance and power efficiency, as well as the size and shape of the devices. In this section of the book, the energy design principles focus is seen in all aspects of technology hardware, from choosing low-power consumption processors and memory to integrating intuitive sensing mechanisms and power management. This is because selecting the appropriate sensor interface protocols and the wireless communication technologies that go with them determines how quickly data is sampled and sent, as well as how complex and power-hungry the process is. In addition, the growing tendency to incorporate intelligence and machine learning into devices is challenging the hardware design to support the required specialized architectures and efficient algorithms that are capable of operating properly within the constraints of these systems. The continuous progress in stretchable electronics alongside eco-materials and energy-efficient AI technology indicates a promising future for wearable sensors with enhanced capabilities and seamless integration in upcoming devices.

Advertisement

3. Non-medical application

Wearable sensors are used in various fields of life and have become an integral part of everyday life. Devices such as smartwatches and fitness trackers are nowadays more sophisticated and finding new uses. Although most often associated with the health industry, these sensors have many other applications besides medical. In this section, the various non-medical applications of wearable sensors are highlighted with examples of how they are used across different sectors and the possible future applications that can be envisioned. Biosensors attached to the skin work by identifying and measuring different biological signals, including heart rate, body temperature or even the electrical activity of the muscles [31]. Signal acquisition relies on sensors converting physiological events into measurable data. For example, heart rate is commonly measured using photoplethysmography (PPG), where light emitted by LEDs reflects off blood vessels under the skin; variations in reflected light intensity, detected by a photodiode, correlate with blood volume changes caused by heartbeats [1, 32]. Muscle activity (electromyography, EMG) is detected via electrodes sensing the electrical potential generated by muscle cells during contraction [2, 3]. This raw sensor data is typically digitized and undergoes initial processing on the device before being transmitted wirelessly, often using Bluetooth Low Energy (BLE), to a smartphone or computer for detailed analysis and visualization [33]. The manner in which these signals are measured depends on the type of sensor used. Some may use biological molecules like enzymes or antibodies to interact with specific substances in body fluids or on the skin. This capacity to capture and analyze biological information is what makes wearable sensors so versatile and amenable to use in a variety of fields.

Despite their use in the management of individual health, research-ready wearable technology offers a unique opportunity for advanced analytics. When incorporated with clinical practice, these sensors can offer valuable information on patient care and outcomes [34]. This points to the wider role of wearable sensors beyond their use in individual disease management to further research and healthcare.

3.1 Wearable sensors in fitness and everyday life

Wearable technology has revolutionized the way people approach fitness and health. Devices like Fitbits and Apple Watches track a number of activities including steps, distance covered, calories consumed, heart rate and even sleep [35]. This data helps people monitor their progress, set goals and make informed decisions about their health. It is important that wearable technology is moving from basic fitness trackers to providing more detailed health information. Some devices, like the Apple Watch Series 10 and Fitbit Sense 4, use AI to check blood oxygen levels, identify heart rhythm irregularities and even predict stress levels [36]. This continuous tracking capability offers users a better understanding of their health than occasional checks and equips them to take precautions for better health. Integration with voice controls and smart home features adds to the list of benefits of wearables in everyday life. Users can control home appliances, unlock cars and get information without having to use their hands, thus making these devices almost indispensable for many.

3.2 Non-medical types of wearable sensors

Wearable devices utilize a variety of sensor technologies to capture and interpret different types of data. Some of the most common types of sensors include (Table 5) [31].

Sensor typeApplicationExample
PressureAltitude monitoring, detecting pressure changesAltimeter watches, fitness bands
HumidityMonitoring local relative humidity, skin moistureMulti-mode watches, fitness bands
PositionMeasuring changes in the angle of a magnetic fieldSmartwatches, fitness trackers
Piezo filmMonitoring user motion and activity levelsFitness bands, motion capture systems
AccelerometerMeasuring acceleration and movementSmartphones, fitness trackers
GyroscopeMeasuring orientation and rotationSmartphones, VR headsets
MagnetometerMeasuring magnetic fields for navigationSmartphones, compass apps
OpticalMeasuring heart rate and blood oxygen levelsSmartwatches, pulse oximeters
TemperatureMeasuring local air temperature or skin temperatureMulti-mode watches, fitness bands

Table 5.

Common sensor types in wearable devices.

3.3 Wearable sensors in sports

Wearable sensors are being used in sports activities and practices, helping athletes and coaches to enhance performance and prevent injuries. Advanced wearables analyze player performance, track biomechanics and even identify early signs of injury [37]. This data helps coaches to design better training programs, manage player workload and improve team performance (Figure 4).

Figure 4.

Illustration of wearable sensors integrated into a sports and fitness context. Examples shown include a smartwatch, chest strap, smart clothing, shoe sensors, earbuds and a sensor hub for data aggregation and transmission.

For example, wearable sensors can measure an athlete’s speed, acceleration and agility. This data can be used to determine where the athlete has potential for improvement and to design specific training plans for the athlete. Wearable sensors can also monitor an athlete’s heart rate and breathing, which can help to prevent overtraining and injuries.

3.4 Wearable sensors in the workplace

Wearable sensors are finding increasing applications in various industries, enhancing safety and productivity. In manufacturing, smart gloves can alert workers to excessive pressure, helping prevent injuries on assembly lines. Wearable sensors can also monitor worker fatigue and stress levels, potentially preventing accidents and promoting a healthier work environment. The resurgence of the Industrial Internet of Things (IoT) is further driving the adoption of wearable sensors in industrial settings [37]. This trend is fueled by the need for improved efficiency, stricter safety regulations and the growing concern for worker safety in increasingly demanding environments. Wearable sensors are being used to monitor worker exposure to hazardous materials, track their location in real-time and provide alerts in case of emergencies. In corporate offices, wearables can encourage employees to take breaks, stand up or engage in physical activity, promoting well-being and reducing burnout. Some companies are even using wearable sensors to track employee productivity and provide feedback on their performance.

3.5 Wearable sensors in entertainment and gaming

Wearable sensors are changing the entertainment industry, especially in gaming and Virtual Reality. Motion-sensing controllers, such as those used in the Nintendo Wii and PlayStation Move, let players interact with the game in a more engaging way [38]. Virtual Reality (VR) headsets work by tracking head and body movements and creating a realistic environment, distinguishing between the real and virtual worlds [39]. Wearable sensors are also being used to create new forms of entertainment. For instance, some companies are developing wearable devices that can be used to track dancers’ movements and give feedback on their performance. This could be used to create interactive dance shows or to help dancers improve their steps.

3.6 Wearable sensors in the movie industry

Wearable sensors are gradually finding their place in the movie industry, especially in the creation of visual effects and animation. Motion capture technology, which uses sensors to replicate an actor’s movements, is currently used to create realistic computer-generated characters [35]. This technology has been used in films such as Avatar and The Lord of the Rings to bring imaginary beings and characters to life. Wearable sensors are also being used to enhance the immersive and interactive movie-watching experience. Some companies are developing products that can be worn to track the audience’s emotions and give feedback to the filmmakers. This could be used to create films that are tailored to the viewer’s preferences.

3.7 Wearable sensors in education

Wearable sensors have the potential to redefine the world of education. Smart glasses can provide students with an immersive learning experience, while sensors integrated into clothing can monitor student engagement and provide feedback to the teacher [40]. This technology can individualize education, inform who needs additional support and create more productive learning environments. For example, wearable sensors can be used to track students’ eye movements and attention spans. This data can be used to identify students who are struggling to focus and who need extra help. Wearable sensors can also be used to measure students’ learning progress and provide them with individualized feedback.

3.8 Wearable sensors in pet care

Wearable sensors are no longer new in the field of pet care; they are gradually becoming popular in improving the quality of life of our best friends. Smart feeders ensure that pets eat the right amount of food at the right time, while GPS trackers provide peace of mind to owners by tracking the location of their pets [41]. Smart pet cameras allow owners to play with and interact with their pets in other ways, thus enhancing the bond between humans and their pets. These sensors can also be used to monitor a pet’s level of activity and duration of sleep, which can give useful information on the animal’s general health. This data can be used to identify potential health issues and decide on the best course of treatment for the animal (Figure 5).

Figure 5.

Wearable sensor applications.

3.9 The future of non-medical wearable sensors

Wearable sensors that are not medical are the future and current and future developments are being made on these devices. The major trends worldwide, including the growth of the digital health sector, extended reality (XR) and the IIoT (Industrial Internet of Things), are greatly impacting the future of wearable sensor development [39]. These trends are leading to the need for smaller, more energy-efficient and more versatile sensors that can be easily integrated into our daily lives. Some of the major trends that will influence the future of wearables are:

  • Miniaturization and integration: Wearable sensors will become less visible, more compact and more diversely incorporated into various objects. This includes making them embedded in clothes and jewelry, and even making them implantable devices [42]. For example, companies like NC State’s ASSIST Center are coming up with “invisible” wearables that can be integrated into clothing or even under the nails [43]. The use of electrochemistry in sensor design also helps in the fabrication of smaller and more flexible devices that can stick to the body [44].

  • Enhanced functionality: Wearables will have better sensors that can tell between various postures and conditions and give more specific information [39]. This includes sensors that can capture a more complete set of biometrics, environmental factors and even emotional states.

  • Improved user experience: Wearables will have better user interfaces and features like voice control, gesture control and haptic feedback for better usability [42]. This will make wearables more convenient and easy to use, and thus available for the use of a wider audience.

  • Energy harvesting: Wearables will use energy harvesting technologies such as solar threads to power themselves and eliminate or reduce the need for batteries [45]. This will improve the convenience and environmental friendliness of the device.

  • Artificial Intelligence (AI): AI will be extremely important in analyzing the large amount of data collected from wearables and presenting the results in a meaningful manner to the users [43]. This will allow wearables to provide more useful information that can help people make decisions about their health, fitness and daily life.

  • Integration with connected vehicles: It is anticipated that wearable sensors will be integrated with connected vehicles to enable cars to check the driver’s alertness, stress levels and even his or her blood alcohol content, which can help prevent accidents and improve the level of driver assistance [39].

Future wearables will not only improve existing technologies but also will be critical in the diagnosis and treatment of disease at an early stage [43]. This could revolutionize healthcare by helping to prevent and improve the accuracy of care.

3.10 Challenges and opportunities

Despite the exciting potential of non-medical wearable sensors, several challenges remain in the field:

  • Data accuracy: This is because the accuracy and precision of the sensor data are important in many applications where accuracy is crucial [46]. Accuracy can be compromised by motion artifacts, signal interference and sensor position variability. Further work is also needed in sensor technology and data processing algorithms.

  • Data security and privacy: Wearables gather a lot of data on people, and this has implications for privacy and security [47]. This paper highlights the need for robust measures to protect this sensitive data. This includes the use of strong encryption, secure storage of the data and clear guidelines on how access to and use of the data will be done.

  • User adoption and engagement: In order to achieve high adoption rates, wearables must be easy to use, comfortable and attractive [46]. This means that design factors, user interfaces and the overall user experience must be taken into consideration.

  • Standardization: The current absence of standards in wearables limits the ability to compare data and share information across different devices [46]. It is important to establish industry standards for data and communication protocols.

  • Technical challenges: There are some technical issues in the development of wearable sensors such as mechanical impedance, noise, delamination and stretching [48]. The human body is a dynamic system that poses several challenges.

This throws up several challenges that new materials, better fabrication techniques and a better understanding of human physiology are needed to address. This is because collaboration between stakeholders is a very vital factor in order to foster innovation, to ensure ethical development of the devices as well as to ensure that they are used properly. Wearable sensors are now available for purposes other than step counting and are changing various aspects of life in various fields including sports, entertainment, education and many other applications.

The use of wearable sensors is increasing and with this use comes societal implications. As these devices gather vast amounts of personal data, it is imperative that ethical considerations are made and principles for development and implementation are needed. The impact of wearables on productivity, privacy, autonomy and equality has to be considered. There are, however, risks and issues that need to be dealt with in order to really realize the opportunities that lie ahead with non-medical wearable sensors.

Advertisement

4. Traditional medical applications

4.1 Traditional medical applications of wearable sensors and devices

Wearable sensors have changed the healthcare sector by giving real-time monitoring and feedback to patients. These devices are embedded with sensors and wireless communication systems that provide real-time data to patients and doctors, which can help in decision-making in the management of the patient. This real-time capability stems from specialized sensors capturing specific biosignals. Electrocardiogram (ECG) sensors, for instance, use skin-contact electrodes to detect the electrical signals generated by the heart’s depolarization and repolarization cycles, providing diagnostic-quality rhythm information [2, 32]. Continuous glucose monitors (CGMs) often utilize minimally invasive electrochemical sensors that measure glucose levels in interstitial fluid via enzymatic reactions (e.g., using glucose oxidase) [4, 49]. The acquired signals are amplified, filtered, digitized by analog-to-digital converters (ADCs) and processed by an onboard microcontroller. Secure data transfer to patient monitoring devices or healthcare platforms is crucial, typically achieved using protocols like BLE for personal devices or sometimes cellular communication for remote patient monitoring systems [3, 6]. Such devices are thought to be crucial for new, digital, preventive and personalized medicine. They are currently used for four main purposes in healthcare:

  • Monitoring: Gathering information from populations with respect to various biophysical processes.

  • Screening: The process of identifying specific conditions and people affiliated with those conditions in a specific dataset.

  • Detection: The use of wearable sensors in diagnosing given biomedical conditions from wearable data.

  • Prediction: The use of wearables in predicting different health states or the progression of a given disease.

This section focuses on conventional medical applications of wearable sensors and devices, including the control of chronic diseases, measurement of vital signs, specific uses in cardiology and neurology and assistance for the elderly. It also describes the drawbacks and promising aspects of this technology (Figure 6).

Figure 6.

Applications of wearable sensors in healthcare. This illustration depicts various uses of wearable sensors, including detecting sleep patterns, measuring respiratory intensity, monitoring blood pressure, detecting vascular infarction and measuring body temperature. The central figure represents the integration of these sensors into a comprehensive health monitoring system.

4.2 Chronic disease management

Wearable sensors are valuable assets in managing chronic diseases, providing real-time and tailored feedback to patients. These devices offer real-time data that patients and healthcare providers can use to monitor health parameters (e.g., heart rate, blood pressure and blood glucose) and trends, enabling adjustments to management plans. Research has shown that wearables enhance treatment outcomes and overall chronic disease management [50]. Furthermore, remote patient monitoring with wearables can improve home care effectiveness and potentially decrease hospital admissions. Wearables also enhance patient engagement in chronic disease care and prevention [51]. This contributes significantly to cost-effectiveness in healthcare [52].

4.3 Diabetes management

Continuous glucose monitors (CGMs) have transformed diabetes management by providing real-time glucose readings without frequent finger-prick tests. These devices use a small skin sensor to measure glucose levels from interstitial fluid, sending data to a receiver like a mobile phone app. CGMs offer numerous advantages:

  • Improved glycemic control: Helps maintain glucose levels within a target range through feedback and alerts [53].

  • Reduced hypoglycemia: Notifies users of low blood sugar risks, enabling preventive action [54].

  • Personalized treatment: Provides data for tailoring insulin regimens and other treatments [49].

Examples include the Dexcom G7 and Abbott Freestyle Libre 2 [55, 56]. Figure 7 illustrates the typical architecture of a CGM system, showing the flow of data from the sensor to the user and healthcare provider.

Figure 7.

A continuous glucose monitoring (CGM) system architecture. The diagram shows the data flow from the diabetic patient’s glucose sensor, through a mobile application and data storage, to the doctor’s phone and monitoring system.

4.4 Cardiovascular disease management

Wearable sensors aid in cardiovascular disease management by tracking heart rate, blood pressure and detecting irregular heartbeats through devices like smartwatches with ECG (Electrocardiogram) capabilities (e.g., Apple Watch) [57]. Wearable blood pressure monitors, such as the Omron HeartGuide, provide continuous monitoring for hypertension control [58].

4.5 Respiratory disease management

Wearable sensors are used in managing respiratory conditions like asthma and COPD by monitoring breathing rate, patterns and oxygen saturation. This continuous tracking helps identify symptom worsening, allowing for early intervention. Data collected by wearables aids healthcare providers in adjusting medications [59].

4.6 Vital signs monitoring

Wearable sensors provide convenient, real-time and potentially continuous measurements of vital signs. Wearable heart rate monitors typically use optical PPG sensors, as described earlier, analyzing light modulation by blood flow [32, 60]. Respiratory rate is often tracked indirectly using accelerometers within IMUs to measure chest wall movements associated with breathing or sometimes estimated from modulations in PPG or ECG signals [3, 61]. Wearable temperature sensors, commonly thermistors, provide real-time skin temperature data, which can indicate trends related to fever or physiological stress [54]. Data from these sensors is usually transmitted wirelessly via BLE [33].

4.7 Types of sensors used in medical wearables

Medical wearable devices monitor various body functions using different sensors, including [62]:

  • Accelerometers: Often part of an inertial measurement unit (IMU), these detect linear acceleration resulting from movement or gravity. The signal originates from the displacement of a proof mass within the sensor. In healthcare, this data helps quantify physical activity levels, analyze gait, monitor sleep posture and detect falls by identifying characteristic impact patterns [3].

  • Heart rate sensors: Primarily use PPG technology, detecting volumetric changes in blood circulation via optical means (light emission and detection) through the skin. This allows continuous monitoring of heart rate and rhythm variability, which is useful in cardiovascular health and fitness assessment [1, 32].

  • ECG sensors: Measure the bioelectrical potentials generated by the heart muscle using conductive electrodes placed on the skin. These provide detailed information about heart rhythm and can detect arrhythmias, which is crucial for diagnosing and managing heart conditions [2].

  • Blood oxygen sensors: Measure peripheral blood oxygen saturation (SpO2) using pulse oximetry. This technique typically uses red and infrared LEDs and a photodetector to measure the differential light absorption characteristics of oxygenated versus deoxygenated hemoglobin pulsating within the arteries [1, 32]. Useful in managing respiratory conditions and assessing overall physiological status.

These integrated sensors provide users and doctors with a comprehensive overview of health status and conditions.

4.8 Kinds of wearable sensors and their applications

Wearable sensors come in various types and are used in different parts of the body, depending on the specific health condition being monitored. Some key types of sensors include [61]:

  • Inertial Measurement Units (IMUs): These sensors, used in fitness trackers, track and estimate data on the user’s sleep, activity and location.

  • Electrochemical biosensors: These sensors monitor chemical levels in the body, such as glucose levels in people with diabetes.

  • Wearable electrodes: These sensors measure electrical activity in the body, such as heart rhythms (ECG) or brain activity (EEG – electroencephalogram).

The table below provides a summary of different sensor types and their applications (Table 6).

Sensor typeApplication
Inertial Measurement Units (IMUs)Monitoring physical activity, sleep patterns and detecting falls
Electrochemical biosensorsMonitoring glucose levels in individuals with diabetes
Wearable electrodesMeasuring heart rhythms (ECG) and brain activity (EEG)

Table 6.

Wearable sensor types and applications.

4.9 Challenges and opportunities

While wearable sensors offer significant potential for improving healthcare, there are also challenges that need to be addressed.

  • Data accuracy and reliability: Correct sensor data is crucial for medical decisions. Potential errors include sensor drift, motion artifacts, skin contact issues and environmental conditions. Addressing these requires sensor calibration, signal processing and validation studies [63].

  • Data security and privacy: Protecting sensitive health data is paramount. This involves strong encryption, data governance policies and informed consent [64].

  • User adoption and adherence: Engaging patients in continuous use is crucial. Factors affecting adherence include device discomfort, ease of use, user motivation, perceived usefulness and integration with daily activities. Approaches include user-centered design and behavioral interventions [65].

  • Interoperability and integration: Integrating wearable data with healthcare systems and EHRs requires data standardization, exchange protocols and consistent security controls among vendors and healthcare organizations [66].

Despite these challenges, the future of wearable sensors in healthcare is promising. Developments in sensor technology, AI and data analysis create opportunities for personalized healthcare, telemedicine and disease prediction [63]. An NIH study demonstrated the potential of smartwatches to indicate inflammation through heart rate changes, opening new possibilities for disease detection [60].

4.10 The future of wearable sensors in healthcare

Wearable sensors are rapidly transforming healthcare, and their impact is expected to grow with advancements in:

  • Miniaturization and comfort: Compact, non-stigmatizing and less intrusive sensors will enhance user adoption, integrating seamlessly into daily life [49].

  • Advanced sensing capabilities: New sensors and materials will enable the monitoring of a wider range of physiological parameters and biomarkers, providing comprehensive health insights [49].

  • AI and machine learning: AI and machine learning algorithms will analyze wearable sensor data, identifying patterns and predicting health outcomes [53].

  • Integration with other technologies: Integration with VR and AR technologies will create immersive healthcare experiences and improve patient engagement [53].

Wearable sensors are poised to transform healthcare by empowering individuals to take control of their health and enabling personalized, proactive care.

Advertisement

5. Latest medical applications

Today, more and more sensors are used in new medical applications. Two recent applications are introduced below.

5.1 Hospital caregivers turning immobile patients

In hospital and long-term care environments, patients often spend extended periods of time lying in bed. After a long time lying in the same position, pressure ulcers can start to form, which may cause pain, irritation, open wounds and infection if left unattended. Pressure ulcers are sometimes referred to as bedsores or pressure sores [67]. As a result, many care homes and hospitals have implemented a policy of turning or rolling a patient in their bed every 2 hours or so. This allows for better blood circulation, decreasing the likelihood of adverse health effects. That being said, turning a patient in their bed can often be a difficult task, and there are usually specific methods used by the caregiver to make sure the patient stays safe throughout the process [68].

This method involves multiple steps. As shown in Figure 8, the first step is that the patient should have their arm folded on shoulder and lie flat on back. The second step is to have the patients’ legs bent slightly and turned halfway with a pillow under their back. The last step would involve the patient being fully turned on their side with a pillow between their thighs [69].

Figure 8.

Turning procedure.

A new application is developed recently in order to assist and analyze the proper patient turning process. The main goal of this project is to develop an automated solution to monitor whether the caregiver is administering care according to the steps shown above [70, 71]. To do this, the researchers used a Raspberry Pi and mounted a camera and other sensors set up with OpenCV and MediaPipe Pose [72]. TensorFlow Lite was also used to make the machine learning model. These three open-source libraries are available in both Python and C++, which makes it more flexible although this project is done with Python. The system allows the camera to recognize specific parts of the body, as well as other objects such as pillows. These can then be used to judge whether or not a patient is in the correct position during the process of being turned. The software can also qualitatively judge each step for correctness and give visible feedback through the form of LED signals.

Figure 9 shows an example of how a person’s posture could be tracked using a set of coordinates on the person’s notable extremities. Once all of the desired points are shown on the screen, MediaPipe Pose will constantly track the person’s position and any changes in it. There are three main steps for turning the patient. The first step involves folding one arm to the opposite shoulder while the other arm remains straight, with both legs being straight during the process. For the second step, the patient should rotate toward the side on which the hand is straight for around 45 ∼ 70 degrees, and both knees should be bent while not being on top of each other. The pillow should be underneath the patient’s back by the end of this step. For the final step, the patient should be fully rotated 90 degrees on their side, their knees should be bent and the neck should remain straight. Knees and ankles should be aligned vertically with the pillow in between.

Figure 9.

Example of posture tracking using coordinates.

If each step is met without error, the green light lights up as an indication to the caregiver that they can move to the next step. To define more rigorously how these steps are met, for the first step, the distance between the palm and the shoulder would be obtained to check if it is lower than the pre-defined distance. The elbow flexing angle will be checked if it is between 35 degrees and 70 degrees to make sure the arm is bending the way it should be. For the arm that should remain straight, it will be checked if it has a flexing angle between 160 and 180 degrees (Figure 10). To take uncertainties in posture into account, a buffer is put in to allow for some degree of error. The last criterion for this step would be checking if the patient’s leg is straight. The program will return a true value if the patient has a hip angle between 70 and 110 degrees and a leg angle of 160 to 180 degrees for both legs.

Figure 10.

Step 1: Prepare to turn.

For the second step, both patient’s knees should be between 50 and 150 degrees (Figure 11). The next criterion for success would be whether the patient is halfway rotated, but there was no way to obtain the angle of the patient if the camera is directly on top of them. To circumvent this, it is required to obtain the distance between both sides of the patient’s shoulders and check whether it is greater than a given distance. The next requirement is that the patient’s legs should not be stacked on one another, so the distance between both sides of the elbows, knees and ankles should be greater than the threshold previously trained, respectively. Lastly, since the pillow needed to be placed behind the patient’s back, the coordinates of the pillows were collected using an object detection algorithm and then checked whether they lay around the back area of the patient. If all the criteria are satisfied, the LED for Step 2 will light up and the software will start detecting criteria for Step 3 (Figure 12).

Figure 11.

Step 2: Half-turning.

Figure 12.

Step 3: Fully turned.

For the last step, the first criterion to check is whether the person is rotated by 90 degrees. This was determined using the distance between both sides of the shoulders. In this case, the distance should be below a threshold. The angle of the knees is also checked to make sure they are bent and their range should be between 90 and 150 degrees. To check whether the patient’s neck is straight, the neck angle is obtained and would only return true if it is between 150 and 210 degrees, such that the patient would have some degree of freedom for their posture while keeping their neck as straight as possible. Lastly, to make sure the pillow is within the patient’s thigh, the coordinates of the pillow are obtained again and the program would check whether the inner corners of the pillow are lying within the patient’s thigh by comparing the coordinates of patient’s knee and heel with the pillow corners.

Advertisement

6. Fall prevention application

Falls are a significant public health concern, especially among seniors and older adults, leading to severe injuries, decreased independence and increased healthcare costs. Automated fall detection systems have emerged as a potential solution to address this issue by detecting falls in real time and alerting caregivers or emergency services. While traditional fall detection methods rely on wearable devices, challenges arise in accurately detecting falls in complex environments. This application proposes a camera-based analysis of human pose using the depth camera, such as Intel RealSense Depth Camera D435 (Figure 13), with the assistance of wearable sensors to enhance the accuracy and reliability of fall detection systems. The system can analyze human poses in three-dimensional space. This depth-based analysis includes calculating the depth of pose landmarks, estimating the body’s center of mass depth and tracking the velocity of movement. The findings of this study contribute to the field of fall detection, offering insights into the advantages of wearable sensors and camera-based analysis of human pose and improving the safety of individuals at risk of falls [73].

Figure 13.

Intel realsense depth camera D435.

The depth camera is utilized in this study, providing synchronized depth and color frames that capture detailed scene depth information. By aligning the depth frames with the corresponding color frames, a robust foundation for analyzing human poses in three-dimensional space is established [74]. The potential of depth information to improve the detection and tracking of human poses is investigated by leveraging the integration of OpenCV, MediaPipe and Numpy libraries in Python.

Depth camera can be used in various situations and applications, such as hand pose tracking in Augmented Reality [74]. To detect and track human poses, a pose module and a depth gradient map module were created, utilizing the depth information to enhance the accuracy of pose estimation. Characteristics, such as the depth of each pose landmark and the approximate depth of the body’s center of mass, are calculated and incorporated into the analysis. The detected poses, including the depth information, coordinates and velocity of selected landmarks, are recorded in a structured manner using a pandas data frame. This enables further analysis and facilitates the future integration of machine learning techniques for fall detection.

A fall detection algorithm is proposed that utilizes the calculated features, such as the depth of specific pose landmarks and the velocity of movement, as inputs to a machine learning model. The features could be classified by the model, which is trained on a labeled dataset of fall and non-fall samples, to detect and differentiate falls from ordinary activities [73].

In order to accurately detect human position, a pose module and a depth gradient map module were created and utilized by the depth-based fall detection system. The capturing of synchronized depth and color frames is done using the RealSense camera, thereby providing comprehensive information for the analysis of human poses in three-dimensional space.

Once the pose landmarks are detected in the color frames, the computation of a depth gradient map using OpenCV takes place, which represents the variations in depth values across the image. By the analysis of the depth gradients around each pose landmark, the approximate depth in the scene is determined. For the calculation of the depth of each pose landmark, consideration is given to the corresponding region in the depth gradient map. The identification of the pixels with high-depth gradients around a particular landmark allows for the inference of its depth relative to the camera. Additionally, the estimation of the approximate depth of the body’s center of mass is achieved by considering the combined information from multiple pose landmarks.

The tracking of the velocity of pose landmarks is performed over consecutive frames in order to capture the dynamics of human movement. The calculation of the displacement of each landmark between frames and the division of it by the time interval allows the velocity vector to be obtained. Fall detection benefits from this velocity information as it enables the analysis of abnormal motion patterns associated with falling.

After recording a video using the D435 camera, individual frames of the video can be manually assigned to specific states based on the pose of the person. This process involves categorizing each frame into one of three distinct states: “safe,” “risky,” and “fallen” (Figure 14). The “safe” state represents the condition in which the person is standing in a normal position, exhibiting no immediate risk of falling over. In this state, the individual’s posture and body alignment indicate a stable and secure stance. The “risky” state corresponds to a precarious situation where the person is at a heightened risk of losing balance and potentially experiencing a fall. Here, the pose characteristics may include factors such as a significant shift in body weight, an unsteady posture or an improper distribution of body mass. Lastly, the “fallen” state denotes the occurrence of an actual fall event, indicating that the person has completely lost balance and has fallen to the ground. The pose in this state typically exhibits distinct patterns, such as a horizontal or near-horizontal alignment of the body.

Figure 14.

Frames are assigned special states.

By systematically assigning frames to these different states, a large dataset comprising a wide range of fall and non-fall samples can be constructed. This dataset serves as a valuable resource for training and evaluating machine learning algorithms designed for fall detection. Each frame provides visual information that can be used to extract meaningful features and patterns indicative of fall events.

This dataset enables the development and refinement of machine learning models that can accurately classify and differentiate between safe, risky and fallen states based on visual cues. By training these models on a diverse set of real-world scenarios, they can learn to generalize and identify potential fall events in various environments and with different individuals.

By employing a camera-based approach for human pose analysis, significant features and patterns that differentiate regular activities from fall events can be extracted. Figure 13 shows the sample of recognition using the trained dataset. Fall detection systems greatly benefit from visual analysis as it plays a vital role in comprehending the dynamics and distinctive attributes of fall events. The technology showcased in this study not only facilitates the creation of a comprehensive dataset encompassing diverse fall and non-fall samples but also reinforces the importance of visual analysis in advancing fall detection methodologies.

The implications of a depth-based analysis of human pose extend beyond fall detection. This technology can be applied to activity recognition, gait analysis, rehabilitation monitoring and elderly care, among others. By understanding and interpreting human pose visually, valuable insights can be extracted to support a range of applications in healthcare, fitness and human-computer interaction (Figure 15).

Figure 15.

System recognizing the state of falling by using trained dataset.

This depth-based analysis of human pose for fall detection presents a promising approach to enhance safety and well-being. By leveraging advanced techniques, such as machine learning, the accuracy, scalability and applicability of fall detection systems can be improved. This research contributes to the growing field of human pose analysis and paves the way for advancements in safety, healthcare and quality of life.

Advertisement

7. Data collection method and machine learning using wearable sensor

In recent years, the use of machine learning has transformed numerous industries and paved the way for groundbreaking innovations. Machine learning enables computer systems to learn from data and make predictions or decisions without being programmed to do so [76]. In the context of wearable sensors, machine learning is primarily employed to automatically recognize patterns within the collected physiological or motion data. These patterns can correspond to specific activities (like walking, running and sleeping), health events (like falls or cardiac arrhythmias) or user states (like stress levels), enabling the device to provide meaningful feedback or alerts [30, 71]. It provides valuable insights that have positioned them as powerful tools for data-driven decision-making and automation.

7.1 Data collection and transmission

Data collection is an essential component of the machine learning algorithms used as predictive models such as patient falls mentioned above [77]. Machine learning algorithms require large data sets to classify patterns related to risky or safe behavior, which can be used to predict and prevent subsequent falls. For the Fall Prevention Project mentioned above, data collection involves tracking the coordinate motion of a watch prototype integrated with various wearable sensors.

The Fall Prevention Project aims to create a device that can help prevent falls by using motion-detecting sensors and depth cameras coupled with machine learning [78]. After data is collected, the device can be utilized to warn caretakers when patients are at risk of falling before it happens. The device fabricated to accomplish this is a wearable watch, composed of a microcontroller attached to an accelerometer and gyroscope sensor. The watch collects coordinate data of the patient’s movements, and from the data collected, a machine learning algorithm is implemented to identify patterns and establish a predictive model.

The sensor used for this project is an MPU6050 accelerometer and gyroscope (Figure 16), a sensor useful for applications that require accurate motion tracking and orientation [78]. The MPU6050 was chosen for its low cost and low power requirements, requiring only 3.3 V of power and 3.6 mA of current while measurements are being taken. The MPU6050 incorporates a gyroscope to measure rotation and angular velocity around an axis and an accelerometer to track changes in velocity and orientation of the device with respect to the earth’s surface. Altogether, this enables the wearer to track complex motion in 3D space. In the context of the Fall Prevention Project, the sensor is principally employed to quantify the position of the wristwatch along the x, y and z axes.

Figure 16.

MPU6050 accelerometer and gyroscope.

The gyroscope relies on the ESP8266 Wi-Fi microchip for data transmission. The Wi-Fi microchip is coded using the Arduino framework to transmit data over the local internet network onto a device such as a laptop. From the user’s device, the data can be formatted, stored and manipulated. The circuitry is integrated with a rechargeable 3.3 V battery into a 3D-printed watch with an adjustable wrist strap designed to accommodate a range of wrist sizes (Figure 17).

Figure 17.

A wrist-attachment prototype watch with multi-sensors.

7.2 Machine learning models

The use of machine learning has accelerated growth in various fields and one of the fields that has benefited the most is medicine. For example, the use of machine learning algorithms to analyze medical data collected by sensors acquired above has produced significant technological advances that have improved patient care and hospital processes [76].

Below four commonly used methods used in machine learning that have proven to be highly effective in a variety of applications will be discussed. By exploring these methods, we gain insight into the diverse techniques employed in machine learning and their practical implementation.

7.2.1 Linear Regression

Linear regression is one of the most commonly understood techniques used for machine learning. A very simple linear regression model consists of one input variable (x) and one output variable (y). By plotting each data point on a set of axes, a “line of best fit” may be drawn through the data (Figure 18). Subsequently, one may predict the value of the output variable (y) for any given input (x). Fundamentally, linear regression seeks to model the linear relationship between a dependent output variable and one or more independent input variables by finding the best-fitting straight line (or hyperplane) through the data points. This method is particularly useful in wearable applications for predicting continuous physiological parameters where a linear relationship with sensor features (like signal characteristics) is assumed or approximated. For example, linear regression models have been developed to estimate blood pressure variations based on features extracted from photoplethysmography (PPG) signals obtained from smartwatches [79].

Figure 18.

Simplified linear regression model.

One particular application of this approach that is more commonly used is least squares regression. Minimizing the sum of the squared residuals (distance from the line of best fit) allows the computer to determine the best internal parameters for the computer to classify future data.

In practice, linear regression usually contains a combination of input variables, which allows for greater complexity in the fitting process for the line of best fit. There are also numerous more optimization methods to improve internal parameters; however, the basis of linear regression remains the same.

7.2.2 QDA

Quadratic discriminant analysis follows a very similar premise to linear regression, with the main difference being the relationship of the input to the output variable. While linear regression assumes a linear relationship between input and output variables, QDA assumes a quadratic relationship. Quadratic discriminant analysis is a probabilistic classification algorithm. It assumes that the measurements from each class (e.g., “walking” and “running”) are drawn from a Gaussian distribution and calculates a quadratic decision boundary between classes. Unlike linear discriminant analysis (LDA), QDA does not assume that the covariance matrices of the classes are equal, allowing for more flexible, curved boundaries. This characteristic makes QDA potentially advantageous for classifying sensor data where the separation between different activities or health states in the feature space is inherently non-linear, such as in complex human activity recognition (HAR) tasks using inertial sensors [80, 81].

More specifically, it assumes a Gaussian distribution of the data and formulates a curve to fit the data as best as possible (Figure 19). As a result, the “line of best fit” starts assuming a “curve of best fit.” One advantage of this approach over linear regression is its precision. For data that do not possess any sort of natural association or contain a lot of outliers, a curve of best fit would be more appropriate to classify it. One downside of it is its computational complexity compared to linear regression.

Figure 19.

Quadratic discriminant analysis (QDA) model.

7.2.3 kNN

The k-Nearest Neighbors algorithm is a technique that uses proximity to determine the classification of a new data point. Selecting the k-nearest neighbors (where k is some constant) and seeing where the larger proportion of neighbors are classified allows the computer to find the appropriate classification for that new data point (Figure 20). The core principle of kNN is “feature similarity”: it classifies a new data point based on the majority class among its “k” nearest neighbors in the multi-dimensional feature space. The “nearness” is typically measured using a distance metric, like Euclidean distance, calculated on features extracted from the sensor data (e.g., mean acceleration, signal variance and frequency components). kNN is relatively simple to implement and is often used as a baseline or effective method for tasks like classifying different types of physical activities (walking, running and cycling) based on patterns observed in accelerometer and gyroscope data collected from a wearable device. Due to its relative simplicity and reliance on feature similarity, kNN is often effective for tasks like classifying different types of physical activities (walking, running and cycling) based on IMU data [82] or distinguishing between sleep and wake states using wrist-worn accelerometer signals [83]. More specifically, the distance to each neighbor from the unclassified data point is stored in a list, which is subsequently sorted in ascending order. Some variations of the algorithm may also assign additional weight to each category based on the relative proximity of each data point to another.

Figure 20.

Simplified k-nearest neighbors (kNN) algorithm.

7.2.4 Unsupervised Learning

Training a “custom model” is also a very widely used technique within machine learning. However, it is notably more abstract than the other three methods outlined above. It involves providing data to the computer with no explicit labels or information about the expected output, which is characterized by the lack of a training set. The computer then performs complex processing tasks to uncover hidden patterns in the data and perform its own categorization (Figure 21). Unlike the supervised methods above, unsupervised learning algorithms work with unlabeled data, seeking to uncover inherent structures or groupings within it. Common techniques include clustering (like k-means), which groups similar data points together and anomaly detection, which identifies data points that deviate significantly from the norm. In wearables, unsupervised learning might be used to automatically discover different types of activities performed by a user without pre-defined labels, cluster users based on their long-term behavior patterns derived from sensor data or detect unusual physiological readings (e.g., an unexpected heart rate spike during rest) that could indicate a potential health issue [84]. Similarly, clustering algorithms can automatically group segments of sensor data corresponding to distinct, potentially unknown, user activities or behavioral states, enabling personalized activity model discovery from long-term wearable recordings [85].

Figure 21.

Unsupervised learning visual aid.

One notable drawback of unsupervised learning is that it is much more computationally intensive compared to supervised learning techniques and usually requires more time to uncover the patterns hidden in the data.

As we move forward, it is crucial to harness the full potential of machine learning while ensuring responsible practices and ethical considerations. By doing so, we can unlock even greater possibilities for innovation, discovery and societal impact, ultimately improving the way we interact with visual data and transforming numerous fields for the better.

Advertisement

8. Future of wearable sensors

Wearable sensors have changed from the basic devices that could measure physical parameters such as steps taken and heart rate to devices that can monitor different aspects of a person’s physical and environmental conditions. This change has been facilitated by the improvement in materials science, microelectronics and data analysis to create new opportunities in healthcare, fitness and interaction between human and machine. This chapter looks at the future of wearable sensors, with emphasis on the present and the emerging trends. This chapter is a review of the current literature on the future of wearable sensors.

8.1 Current trends in wearable sensor technology

The development of wearable sensor technology is still ongoing and more so in terms of making the devices smaller, more accurate and multifunctional. The following table illustrates some of the key trends (Table 7).

TrendDescriptionApplications
Skin-like sensorsResearchers are developing sensors that mimic the properties of human skin, allowing for comfortable and unobtrusive monitoring of various biomarkers. These sensors can conform to the body’s contours and measure temperature, pressure, strain and even biomolecules like glucose and lactate [86].Continuous health monitoring, athletic performance tracking and medical diagnostics
E-tattoosUltrathin, skin-soft electronics that can noninvasively and accurately digitize physiological and psychological information from the human body without compromising comfort or mobility. They can perform a wide range of functions, including sensing, wound healing and energy harvesting [45].Telemedicine, mobile health, human-machine interactions
Smart contact lensesSmart contact lenses are transforming ocular health management and disease monitoring. The integration of sensors and wireless communication within these lenses enables continuous health monitoring [87].Continuous health monitoring, diagnosis and management of ophthalmic, metabolic and neurological diseases
Wearable sweat sensorsWearable sweat sensors continuously and noninvasively monitor health indicators through sweat [88].Personalized health monitoring, precision medicine
Haptic Human-Machine Interfaces (HHMIs)HHMIs enhance how we interact with machines and robots by combining tactile sensation with haptic feedback [89].Immersive experiences in the metaverse, robotics and interactive devices

Table 7.

Current trends in wearable sensor technology.

8.2 Emerging technologies and applications

The following are some of the emerging technologies that are shaping the future of wearable sensors:

  • Artificial Intelligence (AI): AI, particularly machine learning and deep learning models (like convolutional neural networks – CNNs and recurrent neural networks – RNNs/LSTMs), will increasingly analyze the large, complex datasets generated by wearable sensors to provide highly personalized insights, sophisticated pattern recognition and predictive capabilities. These advanced algorithms can automatically learn relevant features from raw time-series sensor data, enabling more accurate detection of subtle health changes (e.g., early disease onset prediction) or nuanced activity classification compared to traditional methods [29, 30]. For instance, AI can be employed in the analysis of sensor data in order to identify the onset of a disease or the likelihood of an injury in athletes.

  • Bioresorbable electronics: These transient electronic systems are able to erode, degrade or cease to exist at certain rates [90] once they are deployed for their purpose. This technology may hold promise to change the current implantable medical devices that require surgical removal. For instance, bioresorbable sensors could be used to assist with the healing process after surgery and then disappear once they are no longer required.

  • Flexible and stretchable electronics: This has only been made possible due to the development of sensor that can be incorporated in clothing, bandages and even implanted in the body [91]. These flexible and stretchable electronics offer new opportunities for conventional and non-invasive health care management. For instance, electronic textiles integrated with sensors can monitor the various vital signs and give instant feedback on physical activities.

  • New materials: The production of wearable sensors has been backed up by the research of gold nanoparticles, graphene pellets that have been treated with antibodies and artificial DNA [92]. These materials can enhance the sensor sensitivity, enabling the detection of a larger number of biomarkers with higher sensitivity.

  • Energy harvesting: This chapter presents a review of the current research into the use of ambient energy sources to power wearable sensors such as body heat, movement and sunlight [93]. This could mean that there will be no need for batteries, hence making the wearable sensors more convenient and sustainable. For instance, mechanical stress of body movement can be employed in the generation of electricity using piezoelectric materials.

These emerging technologies are paving the way for a wide range of new applications for wearable sensors, including:

  • Preventive medicine: By enabling continuous monitoring of various physiological parameters, wearable sensors play a crucial role in preventive medicine, allowing for the early detection of diseases before symptoms become apparent. For example, they can be utilized to identify precursors or early indicators of conditions such as heart disease, diabetes and potentially certain cancers [94]. A specific application involves individuals at risk for cardiac arrhythmias using wearables for extended periods to record infrequent episodes that might be missed during standard clinical evaluations. This capability supports a paradigm shift in healthcare, moving the focus from merely extending “life expectancy” toward enhancing “health expectancy” through proactive health management.

  • Personalized medicine: Wearable sensors can help track people’s response to drugs and treatments and thus help in the development of better care plans and treatment outcomes [95]. Wearable sensors are useful to monitor the levels of drugs in the body and thus help with individualized drug delivery and treatment monitoring.

  • Remote patient monitoring: Wearable sensors can assist healthcare providers in patient care at the patient’s home or in the community, thus reducing the number of hospital admissions [96].

  • Sport and fitness: Wearable sensors can furnish the sportsperson and the fitness enthusiasts with elaborate information about their performance which can go a long way in improving their training and minimizing on injuries. But, it is crucial to consider the negative effect of peer comparison of activity information [97]. While some people may be encouraged by the comparison of their performance with others, others may be demoralized and this may slow them down.

  • Human-computer interaction: Wearable sensors can be used to control other electronic devices for instance smartphones and computers, through gestures and movements [98]. This technology can enhance the accessibility to individuals with disabilities and open new possibilities of human interactions in virtual and augmented reality.

8.3 Challenges and future directions

Despite the rapid progress in wearable sensor technology, several challenges remain:

  • Data security and privacy: Wearable sensors collect sensitive personal data, raising concerns about data security and privacy. Robust security measures and ethical guidelines are needed to protect user data and ensure responsible use of this technology. In addition to security measures, there is a need for expertise in data management and analysis, as well as the development of user-friendly visualization tools to make the data accessible and understandable to both users and healthcare providers [99].

  • User acceptance: The widespread adoption of wearable sensors depends on user acceptance. Devices need to be comfortable, easy to use and provide meaningful insights to encourage long-term use. Concerns about the accuracy of wearables and the complexity of some devices can also influence user adoption. Importantly, user perceptions of wearables are context-dependent and can be influenced by both internal and external factors [100]. Personalized feedback and support from wearables can improve user engagement and long-term adherence.

  • Accessibility: Equitable access to wearable technology is crucial. Concerns about the cost of wearables and the need for smartphones to interpret data can create barriers for some individuals [101]. Addressing these accessibility concerns is essential to ensure that the benefits of wearable sensors are available to everyone.

  • Integration with existing healthcare systems: The integration of wearable sensor data into the current healthcare settings can be quite complex. The need for standardization of data and interchangeable software is therefore important in order to facilitate data collection and analysis. However, it is also important for healthcare providers to be trained and supported to be able to work with the wearable sensor data in clinical practice.

The future of wearable sensors depends on overcoming these challenges and the progress in the field of material science, microelectronics and data analysis. With the continuation of the development of these technologies, wearable sensors will become an inseparable part of people’s lives, changing the field of healthcare, sports and human-machine interfaces. Wearable sensors are not new; we have been using simple wearable devices like pedometers and heart rate monitors for a long time. The development of materials science, microelectronics and data analysis has led to the creation of sophisticated devices that can track numerous parameters of a person’s physiological and environmental state. This technology has the potential to change the fields of healthcare, fitness and human-computer interaction. Some issues that still exist include: security of data, acceptance of the device by the users and the existing healthcare systems. Improvement, therefore, is anticipated in the development of wearable sensor technologies with better technologies, designs and concepts in the future. Wearable sensors can empower people by providing them with individualized support and encouragement, active involvement in the self-care process and better health outcomes [102]. In the future, wearable sensors will become even more widespread and interconnected with people’s lives. In the next update:

  • Increased personalization: Wearable sensors will become increasingly personalized, providing tailored insights and recommendations based on individual needs and preferences.

  • Seamless integration: Wearable sensors will be seamlessly integrated with other technologies, such as smartphones, smart homes and even implanted medical devices.

  • Expanded applications: Wearable sensors will find new applications in areas such as education, workplace safety and environmental monitoring.

Nevertheless, it is crucial to consider the ethical, social and economic implications of this technology. As wearable sensors become more sophisticated and collect more data, issues regarding privacy, data ownership and potential prejudice will be raised. It is crucial to make sure that the creation and application of wearable sensor technology are conducted ethically and in the best interest of people and the society.

Advertisement

9. Conclusions

Wearable sensor technology has been introduced to the public with simple devices like step counters and heart rate monitors. As this chapter shows, they are now fit to monitor and analyze many different types of physiological, environmental and motion-related data that hold a vast potential to transform healthcare, fitness and many other areas of life. This chapter has also established that wearable sensors are helping caregivers in hospitals, preventing falls and improving workplace safety. The integration of machine learning is an additional asset that enhances the capabilities of these devices and has the potential to create even more innovative solutions in the future.

On the other hand, the development of wearable sensor technology is accompanied by several problems. It is important to address data accuracy, security and privacy issues as well as other factors such as user adoption, accessibility and compatibility with existing systems. As we work to overcome these challenges and promote sustainable innovation, wearable sensors can help revolutionize our lives and create a future in which technology is an integral part of our lives.

References

  1. 1. Kwang Bok Kim and Hyun Jae Baek. Photoplethysmography in wearable devices: A comprehensive review of technological advances, current challenges, and future directions. Electronics. 2023;12(13):2923
  2. 2. Sörnmo L, Laguna P. Bioelectrical Signal Processing in Cardiac and Neurological Applications. Burlington, MA, USA: Academic Press; 2005
  3. 3. Lee J, Kim J, Lee J. A survey of wearable sensor systems for healthcare monitoring. Sensors. 2019;19(13):2909
  4. 4. Cappon G, Acciaroli G, Vettoretti M, Facchinetti A, Sparacino G. Wearable continuous glucose monitoring sensors: A revolution in diabetes treatment. Electronics. 2017;6(3):65
  5. 5. Liu Z, Kong J, Menglong Q, Zhao G, Zhang C. Progress in data acquisition of wearable sensors. Biosensors. 2022;12(10):889
  6. 6. Muzny M, Henriksen A, Giordanengo A, Muzik J, Grøttland A, Blixgård H, et al. Wearable sensors with possibilities for data exchange: Analyzing status and needs of different actors in mobile health monitoring systems. International Journal of Medical Informatics. 2020;133:104017
  7. 7. (pdf) Future Trends in Energy-Efficient Wearable Health Monitoring…. Available from: https://www.researchgate.net/publication/388274240_Future_Trends_in_Energy-Efficient_Wearable_Health_Monitoring_Technologies [Accessed: March 17, 2025]
  8. 8. Wireless Communications Technologies for Implementation of Fintech Solutions - African Journals online. Available from: https://www.ajol.info/index.php/iijikm/article/view/286935/270333 [Accessed March 17, 2025]
  9. 9. Sun H, Zhang Z, Rose Qingyang H, Qian Y. Wearable communications in 5g: Challenges and enabling technologies. IEEE Vehicular Technology Magazine. 2018;13(3):100-109
  10. 10. Li Y, Lian Y, Perez V. Design optimization for an 8-bit microcontroller in wireless biomédical sensors. IEEE Biomedical Circuits and Systems Conference. 2009:33-36. DOI: 10.1109/BIOCAS.2009.5372090
  11. 11. Lundager K, Zeinali B, Tohidi M, Madsen JK, Moradi F. Low power design for future wearable and implantable devices. Journal of Low Power Electronics and Applications. 2016;6(4). Article number 20. DOI: 10.3390/jlpea6040020. Available from: https://www.mdpi.com/2079-9268/6/4/20
  12. 12. Samakovlis D, Albini S, Álvarez RR, Constantinescu D-A, Schiavone PD, Peón-Quirós M, et al. Biomedbench: A benchmark suite of tinyml biomedical applications for low-power wearables. IEEE Design & Test. 2024:1-1. DOI: 10.1109/MDAT.2024.3483034
  13. 13. Tsoukas V, Boumpa E, Giannakas G, Kakarountas A. A Review of Machine Learning and Tinyml in Healthcare. PCI ‘21. New York, NY, USA: Association for Computing Machinery; 2022. pp. 69-73. DOI: 10.1145/3503823.3503836
  14. 14. A Taxonomy of Low-Power Techniques in Wearable Medical… - mdpi. Available from: https://www.mdpi.com/2079-9292/13/15/3097 [Accessed: March 17, 2025]
  15. 15. Cortex-m0+ — Processor for Sensors, Wearables, and Low-Power…. Available from: https://www.arm.com/products/silicon-ip-cpu/cortex-m/cortex-m0-plus [Accessed: March 17, 2025]
  16. 16. (pdf) Low-Power Technologies for Wearable Telecare and Telehealth Systems: A review. Available from: https://www.researchgate.net/publication/275253388_Low-power_technologies_for_wearable_telecare_and_telehealth_systems_A_review [Accessed: March 17, 2025]
  17. 17. (pdf) Wearable Processors Architecture: A Comprehensive…. Available from: https://www.researchgate.net/publication/382218319_Wearable_Processors_Architecture_A_Comprehensive_Analysis_of_64-bit_ARM_Processors [Accessed: March 17, 2025]
  18. 18. Zhang K, Zhang CC. Designing a low-cost microcontroller-based rover for microplastic detection using deep-learning image detection and near-infrared spectroscopy. Congress in Computer Science, Computer Engineering, and Applied Computing (CSCE). 2023:957-962. DOI: 10.1109/CSCE60160.2023.00161
  19. 19. Gajaria D, Adegbija T. Evaluating the performance and energy of stt-ram caches for real-world wearable workloads. Future Generation Computer Systems. 2022;136:231-240
  20. 20. Direct Memory Access-Based Data Storage for Long-Term…. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC11315031/ [Accessed: March 17, 2025]
  21. 21. scholarworks.umass.edu. Available from: https://scholarworks.umass.edu/bitstreams/766553c6-0ec8-4f4a-9814-f95c3b30965f/download [Accessed: March 17, 2025]
  22. 22. Energy-Efficient Long-Term Continuous Personal Health Monitoring. Available from: https://oar.princeton.edu/bitstream/88435/pr1639k52q/1/Health_preprint.pdf [Accessed: March 17, 2025]
  23. 23. Energy-Efficient Integration of Continuous Context Sensing and…. Available from: https://www.mdpi.com/1424-8220/15/9/22616 [Accessed: March 17, 2025]
  24. 24. An Efficient Communication Protocol for Real-Time Body Sensor…. Available from: https://www.mdpi.com/2224-2708/14/1/4 [Accessed: March 17, 2025]
  25. 25. An Introduction to i2c and Spi Protocols - Researchgate. Available from: https://www.researchgate.net/publication/224376183_An_introduction_to_I2C_and_SPI_protocols [Accessed: March 17, 2025]
  26. 26. I2c vs Spi: A Comprehensive Comparison and Analysis - Wevolver. Available from: https://www.wevolver.com/article/i2c-vs-spi-protocols-differences-pros-cons-use-cases [Accessed: March 17, 2025]
  27. 27. Wireless Communications for Smart Manufacturing and Industrial…. Available from: https://www.mdpi.com/1424-8220/23/1/73 [Accessed: March 17, 2025]
  28. 28. Power Management for Wearable Electronic Devices — Request pdf. Available from: https://www.researchgate.net/publication/338663018_Power_Management_for_Wearable_Electronic_Devices [Accessed: March 17, 2025]
  29. 29. Advanced Machine Learning and AI Techniques for Enhancing Wearable Health Monitoring Systems - Escholarship. Available from: https://escholarship.org/uc/item/4z89t0h3 [Accessed: March 17, 2025]
  30. 30. Machine Learning for Healthcare Wearable Devices: The Big Picture… Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC9038375/ [Accessed: March 17, 2025]
  31. 31. Roberts R. Wearable biosensors and their applications The Scientist; 2023. Available from: https://www.the-scientist.com/wearable-biosensors-and-their-applications-71360
  32. 32. Pantelopoulos A, Bourbakis NG. A survey on wearable sensor-based systems for health monitoring and prognosis. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews). 2009;40(1):1-12
  33. 33. Mitra S, Chatterjee S, Kumar S. A survey on body sensor networks. Wireless Personal Communications. 2015;82:2549-2573
  34. 34. Casey M et al. Opportunities and challenges for the use of wearable sensors to assess real-world speech in people with Parkinson’s disease. npj Parkinson's Disease. 2023;9(70). Article number 70
  35. 35. Gebbie T. Fitbit vs apple watch: Which is better for you? Livestock Science. 2023. Available from: https://www.livescience.com/fitbit-vs-apple-watch
  36. 36. Romagnoli S, Ripanti F, Morettini M, Burattini L, Sbrollini A. Wearable and portable devices for acquisition of cardiac signals while practicing sport: A scoping review. Sensors. 2023;23(6):3350
  37. 37. Usmani UA, Usmani MU. Future market trends and opportunities for wearable sensor technology. International Journal of Engineering and Technology. 2014;6(4):326
  38. 38. Cao M, Xie T, Chen Z. Wearable Sensors and Equipment in VR Games: A Review. Berlin, Heidelberg: Springer Berlin Heidelberg; 2019. pp. 3-12. DOI: 10.1007/978-3-662-59351-6_1
  39. 39. Shim JP, Varshney U, Dekleva S. Wireless evolution 2006: Cellular TV, wearable computing, and rfid. Communications of the Association for Information Systems. 2006;18(1):24
  40. 40. Khosravi S, Bailey SG, Parvizi H, Ghannam R. Wearable sensors for learning enhancement in higher education. Sensors. 2022;22(19). Article number 7633. DOI: 10.3390/s22197633. Available from: https://www.mdpi.com/1424-8220/22/19/7633
  41. 41. Hussain A, Ali S, Abdullah, Kim H-C. Activity detection for the wellbeing of dogs using wearable sensors based on deep learning. IEEE Access. 2022;10:53153-53163. DOI: 10.1109/ACCESS.2022.3174813
  42. 42. Kim J et al. Wearable sensors: Trends and challenges in 2024. Journal of Wearable Technology. 2024. Available from: https://www.jweartech.com/article/wearable-sensors-trends-2024
  43. 43. Bader A, H, Aarti Rangarajan C, Reddy KK, Rangarajan D, Doss S. A transformative role of wearable health devices from sensors to solutions: An invisible doctor for proactive self-care. In: Utilizing AI of Medical Things for Healthcare Security and Sustainability. Hershey, PA, USA: IGI Global Scientific Publishing; 2025. pp. 277-308
  44. 44. Wang L et al. Flexible electrochemical sensors for wearable applications. Advanced Sensor Materials. 2024. Available from: https://www.advsensormat.com/article/flexible-electrochemical-sensors
  45. 45. Li H, Tan P, Rao Y, Bhattacharya S, Wang Z, Kim S, et al. E-tattoos: Toward functional but imperceptible interfacing with human skin. Chemical Reviews. 2024;124(6):3220-3283
  46. 46. Choi H et al. Accuracy and standardization challenges in wearable sensor data. Wearable Data Journal. 2024. Available from: https://www.wearabledatajournal.com/article/accuracy-standardization-wearable-sensors
  47. 47. Lee S et al. Security and privacy issues in wearable sensor networks. Journal of Network Security. 2024. Available from: https://www.jnetsec.com/article/security-privacy-wearable-sensors
  48. 48. Kim D et al. Stretchable and flexible materials for wearable sensor applications. Advanced Materials Science. 2024. Available from: https://www.advmatsci.com/article/stretchable-flexible-materials-wearable-sensors
  49. 49. Wan S, Li D, Zhou S, Yang S. A survey on wearable sensor-based systems for healthcare monitoring. IEEE Sensors Journal. 2011;11(11):2787-2798
  50. 50. Chen J, Cao Y, Stankovic JA. Data collection and analysis systems in wireless sensor networks: A survey. ACM Transactions on Sensor Networks (TOSN). 2014;10(3):39
  51. 51. Someya T, Sekitani T, Iba S, Kato Y, Kawaguchi H, Sakurai T. A large-area, flexible pressure sensor matrix with organic field-effect transistors for artificial skin applications. Proceedings of the National Academy of Sciences. 2004;101(27):9966-9970
  52. 52. Chong C-Y, Kumar SPS. Sensor networks: Evolution, opportunities, and challenges. Proceedings of the IEEE. 2003;91(8):1247-1256
  53. 53. Chen M, Ma Y, Song J, Lai C-F, Chuan H. Smart clothing: Connecting human with clouds and big data for personalized healthcare. Mobile Networks and Applications. 2016;21(5):825-841
  54. 54. Rogers JA, Someya T, Huang Y. Materials and mechanics for stretchable electronics. Science. 2010;327(5973):1603-1607
  55. 55. Hwang SW, Tao H, Dae-Hyeong Kim H, Cheng J-KS, Rill E, Dobson M, et al. A physically transient form of silicon electronics. Science. 2012;337(6102):1640-1644
  56. 56. Irimia-Vladu M. “Green” electronics: Biodegradable and biocompatible materials and devices for sustainable future. Chemical Society Reviews. 2014;43(2):588-610
  57. 57. Yin L, Huang X, Hai X, Zhang Y, Lam J, Cheng J, et al. Materials, designs, and operational characteristics of mechanically transient electronics. Advanced Materials. 2014;26(28):4745-4752
  58. 58. Anastasi G, Conti M, Di Francesco M, Passarella A. Energy conservation in wireless sensor networks: A survey. Ad Hoc Networks. 2009;7(3):537-568
  59. 59. He Z, Jin L. A comprehensive survey of energy efficient routing protocols in wireless sensor networks. International Journal of Distributed Sensor Networks. 2008;4(2):125-153
  60. 60. Leonov V. Thermoelectric energy harvesting of human body heat for wearable sensors. IEEE Sensors Journal. 2006;6(3):650-657
  61. 61. Paradiso JA, Starner T. Energy scavenging for mobile and wireless electronics. IEEE Pervasive Computing. 2005;4(1):18-27
  62. 62. Starner T. Human-powered wearable computing. IBM Systems Journal. 1996;35(3.4):618-629
  63. 63. Akyildiz IF, Weilian S, Sankarasubramaniam Y, Cayirci E. Wireless sensor networks: A survey. Computer Networks. 2002;38(4):393-422
  64. 64. Gemperle F, Kasabach C, Stivoric J, Bauer M, Martin R. Design for wearability. In: In Digest of Papers. Second International Symposium on Wearable Computers (Cat No. 98EX215). Pittsburgh, PA, USA: IEEE; 1998. pp. 116-122
  65. 65. Moore K, O’Shea E, Kenny L, Barton J, Tedesco S, Sica M, et al. Older adults’ experiences with using wearable devices: Qualitative systematic review and meta-synthesis. JMIR mHealth and uHealth. 2021;9(6):e23832
  66. 66. Gao M, Yao Y, Wang Y, Wang B, Wang P, Wang Y, et al. Wearable power management system enables uninterrupted battery-free data-intensive sensing and transmission. Nano Energy. 2023;107:108107
  67. 67. NHS. Pressure Ulcers (Pressure Sores); 2017. Available from: https://www.nhs.uk/conditions/pressure-sores [Accessed: January 1, 2025]
  68. 68. MedlinePlus. Turning Patients Over in Bed: MedlinePlus Medical Encyclopedia. Available from: https://medlineplus.gov/ency/patientinstructions/000426.htm; 2013 [Accessed: January 1, 2025]
  69. 69. 24 Hour Home Care. Caregiver Training: Turning and Positioning in a Bed - 24 Hour Home Care. In YouTube; 2015. Available from: https://www.youtube.com/watch?v=0R6XoWLUitE [Accessed: January 1, 2025]
  70. 70. Zhang C, C, Shen Y, Gan Z, Wang Y. Using human body recognition to assist hospital caregivers with turning immobile patients. In: 2023 Congress in Computer Science, Computer Engineering, and Applied Computing (CSCE). Las Vegas, NV, USA: IEEE; 2023. pp. 548-552. DOI: 10.1109/CSCE60160.2023.00096
  71. 71. Zhang C, C, Zhang K, Ni R, Liu H, Shen J. Unleashing the potential of machine learning: An exploration of state-of-the-art algorithms and real-world applications in computer vision. In: 2023 Congress in Computer Science, Computer Engineering, and Applied Computing (CSCE). Las Vegas, NV, USA: IEEE; 2023. pp. 422-425. DOI: 10.1109/CSCE60160.2023.00075
  72. 72. Google Developers. Pose Landmarks Detection Task Guide — MediaPipe; 2024. https://developers.google.com/mediapipe/solutions/vision/pose_landmarker [Accessed: January 1, 2025]
  73. 73. Zhang C, C, Wang C, Dai X, Liu S. Camera-based analysis of human pose for fall detection. In: 2023 Congress in Computer Science, Computer Engineering, and Applied Computing (CSCE). Las Vegas, NV, USA: IEEE; 2023. pp. 1779-1782. DOI: 10.1109/CSCE60160.2023.00294
  74. 74. Intel. Intel RealSense Depth Camera D435. Available from: https://www.intelrealsense.com/depth-camera-d435/ [Accessed: January 1, 2025]
  75. 75. Zhang K, Ye MT, Zhang CC, Ni R, Liu Y, Xing A. 3describe-creating tangible AR (augmented reality) objects using depth camera. In: 2023 International Conference on Computational Science and Computational Intelligence (CSCI). Las Vegas, NV, USA: IEEE; 2023. pp. 1280-1283. DOI: 10.1109/CSCI62032.2023.00209
  76. 76. Brown S. Machine Learning, Explained. MIT Sloan, Available from: https://mitsloan.mit.edu/ideas-made-to-matter/machine-learning-explained [Accessed: January 1, 2025]
  77. 77. Lu X, Zhang CC, Wang Y, Shen J, Dai X. Data collection methods and predictive analysis for fall prevention in elderly populations. In: 2023 Congress in Computer Science, Computer Engineering, and Applied Computing (CSCE). Las Vegas, NV, USA: IEEE; 2023. pp. 2740-2742. DOI: 10.1109/CSCE60160.2023.00448
  78. 78. Zhang C, C, Liu MY, Zhang K, Dai X, Wang C, Wang Y. Wearable technology for fall prevention-data collection method and hardware optimization. In: 2023 International Conference on Computational Science and Computational Intelligence (CSCI). Las Vegas, NV, USA: IEEE; 2023. pp. 1498-1502. DOI: 10.1109/CSCI62032.2023.00245
  79. 79. Ren S, Meng W, Zhang X, Yan R, Zhang X, Pan J, et al. Linear regression model for noninvasive blood pressure estimation based on multi-wavelength photoplethysmography. Computers in Biology and Medicine. 2022;146:105530. DOI: 10.1016/j.compbiomed.2022.105530
  80. 80. Uddin MZ, Soylu A. Human activity recognition using wearable sensors, discriminant analysis, and long short-term memory-based neural structured learning. Scientific Reports. 2021;11(1):16455
  81. 81. Leutheuser H, Schuldhaus D, Eskofier BM. Comparative study on human activity recognition with machine learning methods on wearable sensors. In: Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies - Volume 5: HEALTHINF. SCITEPRESS - Science and Technology Publications. San Francisco, CA, USA: Public Library of Science; 2023. pp. 709-716. DOI: 10.5220/0011773700003414
  82. 82. Palimkar P, Bajaj V, Mal AK, Shaw RN, Ghosh A. Unique action identifier by using magnetometer, accelerometer and gyroscope: Knn approach. In: Advanced Computing and Intelligent Technologies: Proceedings of ICACIT 2021. Singapore: Springer; 2022. pp. 607-631
  83. 83. Garcia-Ceja E, Dewri R, Låzăr AS. Comparative analysis of machine learning models for sleep/wake detection using wrist-worn accelerometers. IEEE Journal of Biomedical and Health Informatics. 2022;26(9):4598-4608. DOI: 10.1109/JBHI.2022.3183098
  84. 84. Sunny JS, Pawan C, Patro K, Karnani K, Pingle SC, Lin F, et al. Anomaly detection framework for wearables data: A perspective review on data concepts, data analysis algorithms and prospects. Sensors. 2022;22(3):756
  85. 85. Yazdanian F, Fatehi F, Edirippulige S. Unsupervised Learning Methods for Personalised Human Activity Recognition Using Wearable Devices: A Systematic Review. arXiv preprint arXiv:2305.09435; 2023. Available from: https://arxiv.org/abs/2305.09435
  86. 86. Kim J, Kim J, Kim J. Skin-like sensors: A review. Sensors. 2021;21(19):5638. DOI: 10.3390/s21195638
  87. 87. Han F, Ge P, Wang F, Yang Y, Chen S, Kang J, et al. Smart contact lenses: From rational design strategies to wearable health monitoring. Chemical Engineering Journal. 2024;497:154823
  88. 88. Bariya M, Nyein HYY, Javey A. Wearable sweat sensors. Nature Electronics. 2018;1(3):160-171
  89. 89. Park J, Lee Y, Cho S, Choe A, Yeom J, Ro YG, et al. Soft sensors and actuators for wearable human–machine interfaces. Chemical Reviews. 2024;124(4):1464-1534
  90. 90. Cha GD, Kang D, Lee J, Kim D-H. Bioresorbable electronic implants: History, materials, fabrication, devices, and clinical applications. Advanced Healthcare Materials. 2019;8(11):1801660
  91. 91. Gao W, Ota H, Kiriya D, Takei K, Javey A. Flexible electronics toward wearable sensing. Accounts of Chemical Research. 2019;52(3):523-533
  92. 92. Yi J, Xianyu Y. Gold nanomaterials-implemented wearable sensors for healthcare applications. Advanced Functional Materials. 2022;32(19):2113012
  93. 93. Gao M, Wang P, Jiang L, Wang B, Yao Y, Liu S, et al. Power generation for wearable systems. Energy and Environmental Science. 2021;14(4):2114-2157
  94. 94. Tricoli A, Nasiri N, De S. Wearable and miniaturized sensor technologies for personalized and preventive medicine. Advanced Functional Materials. 2017;27(15):1605271
  95. 95. Tyler J, Choi SW, Tewari M. Real-time, personalized medicine through wearable sensors and dynamic predictive modeling: A new paradigm for clinical medicine. Current Opinion in Systems Biology. 2020;20:17-25
  96. 96. Majumder S, Mondal T, Jamal Deen M. Wearable sensors for remote health monitoring. Sensors. 2017;17(1):130
  97. 97. Aroganam G, Manivannan N, Harrison D. Review on wearable technology sensors used in consumer sport applications. Sensors. 2019;19(9):1983
  98. 98. Yin R, Wang D, Zhao S, Lou Z, Shen G. Wearable sensors-enabled human–machine interaction systems: From design to application. Advanced Functional Materials. 2021;31(11):2008936
  99. 99. Cilliers L. Wearable devices in healthcare: Privacy and information security issues. Health Information Management Journal. 2020;49(2–3):150-156
  100. 100. Wu AY, Munteanu C. Understanding older users’ acceptance of wearable interfaces for sensor-based fall risk assessment. In: Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems. New York, NY, USA. 2018. pp. 1-13
  101. 101. Canali S, Schiaffonati V, Aliverti A. Challenges and recommendations for wearable devices in digital health: Data quality, interoperability, health equity, fairness. PLOS Digital Health. 2022;1(10):e0000104
  102. 102. Sharma A, Singh A, Gupta V, Arya S. Advancements and future prospects of wearable sensing technology for healthcare applications. Sensors and Diagnostics. 2022;1(3):387-404

Written By

Chris Cheng Zhang, Mark Yining Liu, Sky Yangziming Han, Amirreza Sedaghat, Kevin Zhang and Haoyang Wang

Reviewed: 05 May 2025 Published: 14 June 2025