DeepHeart: Accurate Heart Rate Estimation from PPG Signals Based on Deep Learning

Chang, Xiangmao; Li, Gangkai; Tu, Linlin; Xing, Guoliang; Huang, Tian

doi:10.1109/mass.2019.00051

Cited by 15 publications

(9 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In recent years, researchers have tried to address this limitation with data-driven deep learning algorithms. The works of [6], [17] achieve comparable results to the ones of the classical methods, applying a CNN to frequency data and a CNN+LSTM (Long-Short Term Memory) to time data, respectively. In [3], the authors present a CNN architecture which outperforms two classical methods [15], [16] on PPGDalia.…”

Section: Background and Related Work A Temporal Convolutional Networkmentioning

confidence: 89%

Robust and Energy-Efficient PPG-Based Heart-Rate Monitoring

Risso

Burrello

Pagliari

et al. 2021

2021 IEEE International Symposium on Circuits and Systems (ISCAS)

View full text Add to dashboard Cite

A wrist-worn PPG sensor coupled with a lightweight algorithm can run on a MCU to enable non-invasive and comfortable monitoring, but ensuring robust PPG-based heart-rate monitoring in the presence of motion artifacts is still an open challenge. Recent state-of-the-art algorithms combine PPG and inertial signals to mitigate the effect of motion artifacts. However, these approaches suffer from limited generality. Moreover, their deployment on MCU-based edge nodes has not been investigated. In this work, we tackle both the aforementioned problems by proposing the use of hardware-friendly Temporal Convolutional Networks (TCN) for PPG-based heart estimation. Starting from a single "seed" TCN, we leverage an automatic Neural Architecture Search (NAS) approach to derive a rich family of models. Among them, we obtain a TCN that outperforms the previous state-of-theart on the largest PPG dataset available (PPGDalia), achieving a Mean Absolute Error (MAE) of just 3.84 Beats Per Minute (BPM). Furthermore, we tested also a set of smaller yet still accurate (MAE of 5.64 -6.29 BPM) networks that can be deployed on a commercial MCU (STM32L4) which require as few as 5k parameters and reach a latency of 17.1 ms consuming just 0.21 mJ per inference.

show abstract

Section: Background and Related Work A Temporal Convolutional Networkmentioning

confidence: 89%

Robust and Energy-Efficient PPG-Based Heart-Rate Monitoring

Risso

Burrello

Pagliari

et al. 2021

2021 IEEE International Symposium on Circuits and Systems (ISCAS)

View full text Add to dashboard Cite

show abstract

“…ANN generally shows better classification accuracy when training data are given enough, while SVM achieves better classification performance when data are not sufficient (Longjie and Abeysekera, 2019;Liu S. H. et al, 2020). Various types of training data are used to obtain clean PPG features (Chang et al, 2019;Liu X. et al, 2020). For example, interpolated PPG data estimated from ECG by the DeepHeart algorithm and contaminated PPG are provided to train convolution neural network (CNN) for enhancing classification performance.…”

Section: Dependence-based Compensationmentioning

confidence: 99%

Motion Artifact Removal Techniques for Wearable EEG and PPG Sensor Systems

et al. 2021

View full text Add to dashboard Cite

Removal of motion artifacts is a critical challenge, especially in wearable electroencephalography (EEG) and photoplethysmography (PPG) devices that are exposed to daily movements. Recently, the significance of motion artifact removal techniques has increased since EEG-based brain–computer interfaces (BCI) and daily healthcare usage of wearable PPG devices were spotlighted. In this article, the development on EEG and PPG sensor systems is introduced. Then, understanding of motion artifact and its reduction methods implemented by hardware and/or software fashions are reviewed. Various electrode types, analog readout circuits, and signal processing techniques are studied for EEG motion artifact removal. In addition, recent in-ear EEG techniques with motion artifact reduction are also introduced. Furthermore, techniques compensating independent/dependent motion artifacts are presented for PPG.

show abstract

“…This method was used to only detect the MA corrupted signal. To remove the corrupted part of the signal the author Xiangmao Chang et al trained denoising convolutional neural network (DnCNN) with input as clean PPG signal corresponding to PPG signal with MA affected and the clean PPG signal was generated using ECG signal (Chang et al, 2019). The removal of the corrupted part may lead to important data loss.…”

Section: Machine Learning-based Motion Artifacts Reduction Techniquesmentioning

confidence: 99%

“…Wearable health-care systems are also used for the treatment of a few diseases. A wearable device that combines a movable Light Emitting Diode (LED) Ultra Violet (UV) lamp, a camera and an APP is used for the treatment of onychomycosis (Li et al, 2019). A Seizure is an uncontrolled electrical disturbance in the brain which happens suddenly.…”

Section: Movement Disorders Quantificationmentioning

confidence: 99%

Review on the wearable health-care monitoring system with robust motion artifacts reduction techniques

Prabakaran

Rufus

2021

View full text Add to dashboard Cite

Purpose Wearables are gaining prominence in the health-care industry and their use is growing. The elderly and other patients can use these wearables to monitor their vitals at home and have them sent to their doctors for feedback. Many studies are being conducted to improve wearable health-care monitoring systems to obtain clinically relevant diagnoses. The accuracy of this system is limited by several challenges, such as motion artifacts (MA), power line interference, false detection and acquiring vitals using dry electrodes. This paper aims to focus on wearable health-care monitoring systems in the literature and provides the effect of MA on the wearable system. Also presents the problems faced while tracking the vitals of users. Design/methodology/approach MA is a major concern and certainly needs to be suppressed. An analysis of the causes and effects of MA on wearable monitoring systems is conducted. Also, a study from the literature on motion artifact detection and reduction is carried out and presented here. The benefits of a machine learning algorithm in a wearable monitoring system are also presented. Finally, distinct applications of the wearable monitoring system have been explored. Findings According to the study reduction of MA and multiple sensor data fusion increases the accuracy of wearable monitoring systems. Originality/value This study also presents the outlines of design modification of dry/non-contact electrodes to minimize the MA. Also, discussed few approaches to design an efficient wearable health-care monitoring system.

show abstract

DeepHeart: Accurate Heart Rate Estimation from PPG Signals Based on Deep Learning

Cited by 15 publications

References 22 publications

Robust and Energy-Efficient PPG-Based Heart-Rate Monitoring

Robust and Energy-Efficient PPG-Based Heart-Rate Monitoring

Motion Artifact Removal Techniques for Wearable EEG and PPG Sensor Systems

Review on the wearable health-care monitoring system with robust motion artifacts reduction techniques

Contact Info

Product

Resources

About