Deep Learning for Heart Rate Estimation From Reflectance Photoplethysmography With Acceleration Power Spectrum and Acceleration Intensity

Abstract

A wearable reflectance-type photoplethysmography (PPG) sensor can be incorporated in a watch or band to provide instantaneous heart rates (HRs) with minimum inconvenience to users. However, the sensor is sensitive to motion artifacts (MAs), which results in inaccurate HR estimation. To address this problem, we propose a new neural network for deep learning to ensure accurate HR estimation even during intensive exercise. Methods: We propose a new deep neural network based on multiclass and non-uniform multilabel classification for HR estimation. It comprises of two convolutional layers, two long short-term memory (LSTM) layers, one concatenation layer, and three fully connected layers including a softmax. The proposed model feeds the power spectra from the PPG and acceleration signals along with the acceleration intensity to the input layer. We also present a new scheme to evaluate the loss value by modifying the true HR value into a Gaussian distribution. Results: We used 48 training datasets and evaluated 23 isolated testing datasets. The proposed model exhibited average absolute error of less than 1.5 bpm for all the training and test datasets-1.09 bpm for the training dataset and 1.46 bpm for the test dataset. Conclusion: The proposed model outperforms the state-of-the-art methods for accurate estimation of HR. Significance: It precisely estimates the HRs with robustness even during intensive physical exercise, as evidenced by the accuracy when PPG signals are severely corrupted by MAs. INDEX TERMS Reflectance-type photoplethysmography, instantaneous heart rate, deep learning, convolutional layer, long short-term memory (LSTM) layer.

A Review of Wearable Multi-Wavelength Photoplethysmography

“…Using IR light as a motion reference [116] allows for a more efficient solution as motion sensors such as accelerometers are not required as well as algorithmic approaches using several wavelengths [72]. Research in single-wavelength PPG sensing has explored the use of machine learning and deep learning models, such as Convolutional Neural Networks and Long Short-Term Memory networks, to accurately and robustly remove motion artifacts and estimate heart rate [128]- [130] and blood pressure [131]. This methodology would be well suited to multi-wavelength PPG sensing and should be explored further.…”

Multi-Sensor Wearable Health Device Framework for Real-Time Monitoring of Elderly Patients Using a Mobile Application and High-Resolution Parameter Estimation

“…In Puranik and Morales ( 2020 ), a digital filter for PPG signals collected from an MWHD is proposed using an adaptive neural network, allowing a more accurate estimate of the HR, resulting in a variation of 3% concerning the ground truth. In Chung et al ( 2020 ), a deep learning approach is proposed for the HR estimation using PPG signals, achieving an absolute error of 1.5 beats-per-minute (BPM), outperforming state-of-the-art methods. Finally, the authors of Panwar et al ( 2020 ) present a new deep learning model with the capability to estimate HR using only a single channel provided by the PPG signal, achieving an error for HR estimation of 0.046 BPM.…”

A Real-Time PPG Peak Detection Method for Accurate Determination of Heart Rate during Sinus Rhythm and Cardiac Arrhythmia

“… Our proposed algorithms showed reliable results not only for NSR data, but also for various cardiac arrhythmias such as PAC/PVC, basal AF, and AF with RVR. Unlike deep learning algorithms that only track heart rates on NSR subjects [ 15 , 16 , 17 ], our algorithm is computationally efficient and was embedded in Samsung Gear S3 and Galaxy Watch 3 smartwatches for heart rate estimation and AF detection in an NIH-funded clinical trial (study ID NCT03761394). …”

“…Unlike deep learning algorithms that only track heart rates on NSR subjects [ 15 , 16 , 17 ], our algorithm is computationally efficient and was embedded in Samsung Gear S3 and Galaxy Watch 3 smartwatches for heart rate estimation and AF detection in an NIH-funded clinical trial (study ID NCT03761394).…”