Physics Maths Engineering
An energy-efficient semi-supervised approach for on-device photoplethysmogram signal quality assessment
Mohammad Feli,
Iman Azimi,
Arman Anzanpour,
Amir M. Rahmani
Peer Reviewed
Abstract
Photoplethysmography (PPG) is a non-invasive technique used in wearable devices to measure vital signs (e.g., heart rate). The method is, however, highly susceptible to motion artifacts, which are inevitable in remote health monitoring. Noise reduces signal quality, leading to inaccurate decision-making. In addition, unreliable data collection and transmission waste a massive amount of energy on battery-powered devices. Studies in the literature have proposed PPG signal quality assessment (SQA) enabled by rule-based and machine learning (ML)-based methods. However, rule-based techniques were designed according to certain specifications, resulting in lower accuracy with unseen noise and artifacts. ML methods have mainly been developed to ensure high accuracy without considering execution time and device’s energy consumption. In this paper, we propose a lightweight and energy-efficient PPG SQA method enabled by a semi-supervised learning strategy for edge devices. We first extract a wide range of features from PPG and then select the best features in terms of accuracy and latency. Second, we train a one-class support vector machine model to classify PPG signals into “Reliable” and “Unreliable” classes. We evaluate the proposed method in terms of accuracy, execution time, and energy consumption on two embedded devices, in comparison to five state-of-the-art PPG SQA methods. The methods are assessed using a PPG dataset collected via smartwatches from 46 individuals in free-living conditions. The proposed method outperforms the other methods by achieving an accuracy of 0.97 and a false positive rate of 0.01. It also provides the lowest latency and energy consumption compared to other ML-based methods.
Key Questions
What challenges are associated with using Photoplethysmography (PPG) in wearable devices?
PPG is a non-invasive method commonly used in wearable devices to monitor vital signs like heart rate. However, it is highly susceptible to motion artifacts, which can degrade signal quality, lead to inaccurate health assessments, and increase energy consumption due to unreliable data collection and transmission.
What methods have been proposed to assess PPG signal quality, and what are their limitations?
Existing methods for PPG signal quality assessment (SQA) include rule-based and machine learning (ML)-based approaches. Rule-based methods are designed according to specific criteria, resulting in lower accuracy when encountering unforeseen noise and artifacts. ML-based methods often achieve high accuracy but may not consider execution time and energy consumption, making them less suitable for resource-constrained wearable devices.
What is the proposed solution for improving PPG signal quality assessment in edge devices?
The study proposes a lightweight and energy-efficient PPG SQA method enabled by a semi-supervised learning strategy for edge devices. It involves extracting a wide range of features from PPG signals, selecting the most effective features in terms of accuracy and latency, and training a one-class support vector machine model to classify PPG signals into "Reliable" and "Unreliable" categories.
How does the proposed method perform compared to existing PPG SQA methods?
The proposed method outperforms five state-of-the-art PPG SQA methods, achieving an accuracy of 0.97 and a false positive rate of 0.01. It also provides the lowest latency and energy consumption compared to other ML-based methods, making it well-suited for implementation in wearable devices.
What dataset was used to evaluate the proposed PPG SQA method?
The evaluation was conducted using a PPG dataset collected via smartwatches from 46 individuals in free-living conditions, ensuring the method's effectiveness in real-world scenarios.
