Abstract:Recent studies showed that Photoplethysmography (PPG) sensors embedded in wearable devices can estimate heart rate (HR) with high accuracy. However, despite of prior research efforts, applying PPG sensor based HR estimation to embedded devices still faces challenges due to the energy-intensive high-frequency PPG sampling and the resource-intensive machine-learning models. In this work, we aim to explore HR estimation techniques that are more suitable for lower-power and resource-constrained embedded devices. More specifically, we seek to design techniques that could provide high-accuracy HR estimation with low-frequency PPG sampling, small model size, and fast inference time. First, we show that by combining signal processing and ML, it is possible to reduce the PPG sampling frequency from 125 Hz to only 25 Hz while providing higher HR estimation accuracy. This combination also helps to reduce the ML model feature size, leading to smaller models. Additionally, we present a comprehensive analysis on different ML models and feature sizes to compare their accuracy, model size, and inference time. The models explored include Decision Tree (DT), Random Forest (RF), K-nearest neighbor (KNN), Support vector machines (SVM), and Multi-layer perceptron (MLP). Experiments were conducted using both a widely-utilized dataset and our self-collected dataset. The experimental results show that our method by combining signal processing and ML had only 5% error for HR estimation using low-frequency PPG data. Moreover, our analysis showed that DT models with 10 to 20 input features usually have good accuracy, while are several magnitude smaller in model sizes and faster in inference time.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is: how to achieve efficient and accurate heart rate estimation on resource - constrained embedded devices. Specifically, the paper focuses on how to improve the efficiency and accuracy of heart rate estimation by reducing the sampling frequency of PPG sensors, decreasing the size of machine - learning models, and shortening the inference time, thus making these techniques more suitable for use in low - power and resource - constrained devices. ### Problems Solved in the Paper: 1. **Energy Consumption Problem Caused by High Sampling Frequency**: - Traditional heart rate estimation methods based on PPG signals usually require a relatively high sampling frequency (such as 125 Hz), which will lead to high energy consumption and is not suitable for resource - constrained embedded devices. - The paper proposes to reduce the sampling frequency from 125 Hz to 25 Hz while maintaining or even improving the accuracy of heart rate estimation by combining signal - processing and machine - learning methods. 2. **Resource Consumption Problem of Machine - Learning Models**: - Existing machine - learning - based heart rate estimation methods usually require large models and long inference times, which are difficult to implement on resource - constrained embedded devices. - By exploring different machine - learning models (such as decision trees, random forests, K - nearest neighbors, support vector machines, and multi - layer perceptrons), the paper finds solutions with smaller model sizes and shorter inference times while ensuring accuracy. 3. **Limitations of Data Sets**: - Existing data sets (such as the ISPC data set) are usually short and not sufficient for extensive evaluation. - The paper collects a new data set, which contains PPG signals and electrocardiogram data as long as 2 hours, for more comprehensive evaluation. ### Solutions: 1. **Combining Signal Processing and Machine Learning**: - First, generate a rough heart rate estimate through signal processing. - Then, input these rough heart rate estimates into a smaller machine - learning model to further improve the accuracy of heart rate estimation. 2. **System Architecture**: - **Stage 1**: PPG data collection. - **Stage 2**: Signal - processing stage, including initial heart rate extraction, Z - score outlier filtering, and smoothing processing. - **Stage 3**: Machine - learning stage, using serialized rough heart rate estimates to generate more accurate heart rate estimates. - **Stage 4**: Reporting module, reporting the estimated heart rate results to the user. 3. **Experimental Evaluation**: - Use the ISPC data set and the self - collected data set for experiments. - Evaluate the accuracy, model size, and inference time of different machine - learning models (DT, RF, KNN, SVM, MLP) under different numbers of features. - The experimental results show that the decision - tree (DT) model usually has higher accuracy, a smaller model size, and a shorter inference time when there are 10 to 20 input features. ### Conclusion: The paper successfully achieves high - precision heart rate estimation at a low sampling frequency (25 Hz) by combining signal - processing and machine - learning methods, and finds a machine - learning model configuration suitable for resource - constrained devices. This method not only reduces energy consumption but also improves the efficiency and accuracy of heart rate estimation.

PPG-based Heart Rate Estimation with Efficient Sensor Sampling and Learning Models

Heart Rate Estimation Using Wrist-Acquired Photoplethysmography under Different Types of Daily Life Motion Artifact.

Efficient and Direct Inference of Heart Rate Variability using Both Signal Processing and Machine Learning

Real-Time Heart Rate Estimation Algorithm Based on Adaptive Spectrum Correction and Peak Localization

Embedding Temporal Convolutional Networks for Energy-efficient PPG-based Heart Rate Monitoring

Edge-Enabled Heart Rate Estimation from Multisensor PPG Signals

Q-PPG: Energy-Efficient PPG-based Heart Rate Monitoring on Wearable Devices

Tiny-HR: Towards an interpretable machine learning pipeline for heart rate estimation on edge devices

PPGnet: Deep Network for Device Independent Heart Rate Estimation from Photoplethysmogram

A novel motion and Noise Artifacts Reduction Mechanism (MNARM) for wearable PPG-based heart rate extraction

An effective photoplethysmography heart rate estimation framework integrating two-level denoising method and heart rate tracking algorithm guided by finite state machine

Variational Mode Decomposition-Based Heart Rate Estimation Using Wrist-Type Photoplethysmography During Physical Exercise

PPG-Based Smart Wearable Device With Energy-Efficient Computing for Mobile Health-Care Applications

An Ultralow-Power PPG Sensor with Adaptive Predictive Sampling

A New Approach for Heart Rate Monitoring Using Photoplethysmography Signals Contaminated by Motion Artifacts

High-accuracy heart rate detection using multispectral IPPG technology combined with a deep learning algorithm

A solution for co-frequency and low SNR problems in heart rate estimation based on photoplethysmography signals

Removal Of Motion Artifacts In Photoplethysmograph Sensors During Intensive Exercise For Accurate Heart Rate Calculation Based On Frequency Estimation And Notch Filtering

Lightweight and interpretable convolutional neural network for real-time heart rate monitoring using low-cost video camera under realistic conditions

Multi-Head Cross-Attentional PPG and Motion Signal Fusion for Heart Rate Estimation

Robust Heart Rate Monitoring for Quasi-Periodic Motions by Wrist-Type PPG Signals