skip to main content
Explore Scholarly Publications and Datasets in the NSF-PAR

Title: An Accurate Non-accelerometer-based PPG Motion Artifact Removal Technique using CycleGAN
A photoplethysmography (PPG) is an uncomplicated and inexpensive optical technique widely used in the healthcare domain to extract valuable health-related information, e.g., heart rate variability, blood pressure, and respiration rate. PPG signals can easily be collected continuously and remotely using portable wearable devices. However, these measuring devices are vulnerable to motion artifacts caused by daily life activities. The most common ways to eliminate motion artifacts use extra accelerometer sensors, which suffer from two limitations: i) high power consumption and ii) the need to integrate an accelerometer sensor in a wearable device (which is not required in certain wearables). This paper proposes a low-power non-accelerometer-based PPG motion artifacts removal method outperforming the accuracy of the existing methods. We use Cycle Generative Adversarial Network to reconstruct clean PPG signals from noisy PPG signals. Our novel machine-learning-based technique achieves 9.5 times improvement in motion artifact removal compared to the state-of-the-art without using extra sensors such as an accelerometer, which leads to 45% improvement in energy efficiency.
Authors:
; ; ; ;
Award ID(s):
2028782
Publication Date:
NSF-PAR ID:
10359654
Journal Name:
ACM Transactions on Computing for Healthcare
ISSN:
2691-1957
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ( , SPIE Defense + Commercial Sensing)
    Dutta, Achyut K. ; Balaya, Palani ; Xu, Sheng (Ed.)
    Monitoring human health in real-time using wearable and implantable electronics (WIE) has become one of the most promising and rapidly growing technologies in the healthcare industry. In general, these electronics are powered by batteries that require periodic replacement and maintenance over their lifetime. To prolong the operation of these electronics, they should have zero post-installation maintenance. On this front, various energy harvesting technologies to generate electrical energy from ambient energy sources have been researched. Many energy harvesters currently available are limited by their power output and energy densities. With the miniaturization of wearable and implantable electronics, the size of the harvesters must be miniaturized accordingly in order to increase the energy density of the harvesters. Additionally, many of the energy harvesters also suffer from limited operational parameters such as resonance frequency and variable input signals. In this work, low frequency motion energy harvesting based on reverse electrowetting-ondielectric (REWOD) is examined using perforated high surface area electrodes with 38 µm pore diameters. Total available surface area per planar area was 8.36 cm2 showing a significant surface area enhancement from planar to porous electrodes and proportional increase in AC voltage density from our previous work. In REWOD energy harvesting, high surface areamore »electrodes significantly increase the capacitance and hence the power density. An AC peak-to-peak voltage generation from the electrode in the range from 1.57-3.32 V for the given frequency range of 1-5 Hz with 0.5 Hz step is demonstrated. In addition, the unconditioned power generated from the harvester is converted to a DC power using a commercial off-theshelf Schottky diode-based voltage multiplier and low dropout regulator (LDO) such that the sensors that use this technology could be fully self-powered. The produced charge is then converted to a proportional voltage by using a commercial charge amplifier to record the features of the motion activities. A transceiver radio is also used to transmit the digitized data from the amplifier and the built-in analog-to-digital converter (ADC) in the micro-controller. This paper proposes the energy harvester acting as a self-powered motion sensor for different physical activities for wearable and wireless healthcare devices.« less
  2. ; ; ( , 236th Electrochemical Society Meeting Abstracts)
    Force sensors play an important role in the biomedical devices industry, especially in motion- and pressure-related devices. Such sensors are designed to collect force or pressure data by converting it into electrical signals. The data can then be sent to and analyzed by a local or cloud-based processing unit. It is vital that the sensors can be fabricated in a way that time efficiency, cost efficiency, and quality are all maximized. The advent of three-dimensional (3D) printing has greatly facilitated prototyping and customized manufacturing, as compared to older crafting methods (such as welding and woodworking), 3D printing requires less skill and involves less costly materials making it much more time- and cost-efficient. Technological advancements have also improved the quality of the actual sensing materials used in sensor-based devices, and notably, carbon-based materials have become increasingly favored for use as sensing elements. In the presented sensor, the modern sensor fabrication methods of 3D printing and using carbon materials as sensing elements are combined. The sensor presented as a proof of the above concepts is a cantilever flex sensor. The sensor consists of a 30 mm-long cantilever extending from a 2.5 mm-thick wall, with a second wall of the same thickness parallelmore »to the cantilever. After designing this structure and printing it using a 3D printer, the top surface of the cantilever was coated with a thin layer of conductive carbon paste and two copper wires were stripped and soldered to a pair of copper alligator clips, to be used for testing purposes. To test the sensor, the two copper wires were clipped onto the sensor (Figure 1A) and each wire was connected to a multimeter probe on the end opposite of the alligator clip. Then, using a set of four through holes in the parallel wall (along with a slotted rod), the tip of the cantilever was pressed down to an angle of 5, 10, 15, or 20 degrees (Figures 1B, 1C, 1D, and 1E, respectively) below the original plane of the cantilever and held there for 2 minutes. The resistance between the ends of the cantilever was measured throughout each trial by the multimeter, and the results (Figure 1F) for each angle were compiled and analyzed to determine the effect of each depression angle on impedance change, and thus, the overall effectiveness of the sensor. In the future, a notable improvement would be miniaturizing the sensor to facilitate in integration of the sensor in wearable and biomedical devices.« less
  3. ; ; ; ; ( , 2019 IEEE International Conference on Embedded Software and Systems (ICESS))
    Advances in integrated sensors and low-power electronics have led to an increase in the use of wearable devices for health and activity monitoring applications. These devices have severe limitations on weight, form-factor, and battery size since they have to be comfortable to wear. Therefore, they must minimize the total platform energy consumption while satisfying functionality (e.g., accuracy) and performance requirements. Optimizing the platform-level energy efficiency requires considering both the sensor and processing subsystems. To this end, this paper presents a sensor-classifier co-optimization technique with human activity recognition as a driver application. The proposed technique dynamically powers down the accelerometer sensors and controls their sampling rate as a function of the user activity. It leads to a 49% reduction in total platform energy consumption with less than 1% decrease in activity recognition accuracy.
  4. ; ; ; ; ; ; ( , JMIR mHealth and uHealth)
    Background Research has shown the feasibility of human activity recognition using wearable accelerometer devices. Different studies have used varying numbers and placements for data collection using sensors. Objective This study aims to compare accuracy performance between multiple and variable placements of accelerometer devices in categorizing the type of physical activity and corresponding energy expenditure in older adults. Methods In total, 93 participants (mean age 72.2 years, SD 7.1) completed a total of 32 activities of daily life in a laboratory setting. Activities were classified as sedentary versus nonsedentary, locomotion versus nonlocomotion, and lifestyle versus nonlifestyle activities (eg, leisure walk vs computer work). A portable metabolic unit was worn during each activity to measure metabolic equivalents (METs). Accelerometers were placed on 5 different body positions: wrist, hip, ankle, upper arm, and thigh. Accelerometer data from each body position and combinations of positions were used to develop random forest models to assess activity category recognition accuracy and MET estimation. Results Model performance for both MET estimation and activity category recognition were strengthened with the use of additional accelerometer devices. However, a single accelerometer on the ankle, upper arm, hip, thigh, or wrist had only a 0.03-0.09 MET increase in prediction error comparedmore »with wearing all 5 devices. Balanced accuracy showed similar trends with slight decreases in balanced accuracy for the detection of locomotion (balanced accuracy decrease range 0-0.01), sedentary (balanced accuracy decrease range 0.05-0.13), and lifestyle activities (balanced accuracy decrease range 0.04-0.08) compared with all 5 placements. The accuracy of recognizing activity categories increased with additional placements (accuracy decrease range 0.15-0.29). Notably, the hip was the best single body position for MET estimation and activity category recognition. Conclusions Additional accelerometer devices slightly enhance activity recognition accuracy and MET estimation in older adults. However, given the extra burden of wearing additional devices, single accelerometers with appropriate placement appear to be sufficient for estimating energy expenditure and activity category recognition in older adults.« less
  5. ; ; ( , 2019 IEEE International Symposium on Smart Electronic Systems (iSES) (Formerly iNiS))
    Overindulgence of harmful substances such as drugs or alcohol, called substance abuse, can directly affect a person's health and their day-to-day activities. The younger population become more vulnerable to such use of psychoactive substances due to lack of awareness of the long-term hazardous effects these substances can have on their health. Additionally, these individuals tend to develop severe mental disorders as they grow older. With the boom of Internet of Things (IoT), the use of wearable sensors such as smartwatches and smartphones has tremendously increased. These wearables help in monitoring a person's physiological signal and keep them informed of one's health. In this research, we propose an edge-intelligent IoT-based wearable that can assist in substance-abuse detection by monitoring their physiological signals on daily basis. The proposed system helps in monitoring the substance abuse and craving of the individual and help the healthcare provider to start an early intervention as required. The proposed system is validated using a custom-built wearable, i-SAD, which was developed as a dedicated substance abuse wearable using commercially available off-the-shelf components. The proposed wearable design was validated using medical quality wearable and yielded a correlation of 0.89 for accelerometer values and 0.92 for average heart rate values.
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email