Key result
A fully binarized network (bCorNET) for heart rate estimation from wrist-PPG achieved a mean absolute error of 6.67 ± 5.49 bpm while consuming 56.1 μJ per window.
Why the study?
Existing deep learning networks for HR estimation from motion-artifact-induced wrist-PPG signals have large numbers of parameters, presenting challenges for low-complexity hardware implementation and on-node processing.
A fully binarized neural network for heart rate estimation from wrist-PPG signals demonstrates low energy consumption and acceptable accuracy, enabling efficient on-node processing for wearable devices.
May enable low-power on-device HR estimation in wearables; leaves open clinical validation before practice adoption.
Research on heart rate (HR) estimation using wrist-worn photoplethysmography (PPG) sensors have progressed rapidly owing to the prominence of commercial sensing modules, used widely for lifestyle monitoring. Reported methodologies have been fairly successful in mitigating the effect of motion artifacts (MA) in ambulatory environment for HR estimation. Recently, a learning framework, CorNET, employing two-layer convolution neural networks (CNN) and two-layer long short-term network (LSTM) was successfully reported for estimating HR from MA-induced PPG signals. However, such a network topology with large number of parameters presents a challenge, towards low-complexity hardware implementation aimed at on-node processing. In this paper, we demonstrate a fully binarized network (bCorNET) topology and its corresponding algorithm-to-architecture mapping and energy-efficient implementation for HR estimation. The proposed framework achieves a MAE of 6.67 ± 5.49 bpm when evaluated on 22 IEEE SPC subjects. The design, synthesized with ST65 nm technology library achieving 3 GOPS @ 1 MHz, consumes 56.1 μJ per window with occupied 1634K NAND2 equivalent cell area and had a latency of 32 ms when estimating HR every 2 s from PPG signals.
No takes yet. Share an insight, caveat, or question.
Rocha et al. (2020) studied Heart rate estimation (n=22). Fully binarized network (bCorNET) topology was evaluated on Mean absolute error (MAE) of heart rate estimation. A fully binarized network (bCorNET) for heart rate estimation from wrist-PPG achieved a mean absolute error of 6.67 ± 5.49 bpm while consuming 56.1 μJ per window.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: