Development of deep learning models for motion artifact mitigation in wearable PPG devices
Bibliographic record
Abstract
Wearable devices are becoming more significant in remote healthcare, fitness tracking, and athletics monitoring through advancing sensor technologies. Primarily utilizing photoplethysmography, these devices employ light to non-invasively monitor physiological changes in peripheral circulation, allowing for the estimation of heart rate, pulse oximetry, blood flow, arrhythmia detection, and other vital cardiac metrics. A significant challenge in the utility of these optical signals is the presence of motion artifacts, which diminish the usefulness of data from wearable devices during exercise and high-movement activities. Currently, accelerometers detect excessive motion and dispose of ‘contaminated’ data. Despite this, accelerometers capture global motion rather than the relative motion at the sensor-skin interface, which is the predominant source of motion artifacts. Here, we propose integrating a pressure channel and multiple light wavelengths within the wearable to support the denoising process. The pressure channel captures relative motion, and an increased spectral resolution captures blood flow information at different depths. This multimodal data can be combined with modern deep learning models to reconstruct motion-free physiological waveforms. We developed a multisensory device that combines force and multiwavelength optical measurements to capture relative motion at the sensor interface. Deep learning models were then trained to reconstruct the photoplethysmography waveform. Initial testing and training consist of controlled lab experiments with plans to translate into real-world examples of motion. Current bench testing has resulted in a model that reconstructs denoised PPG signals with a leave-one-subject-out mean square error of 0.037, mean heart rate estimation error of 1.35 BPM, and an overall 23.68% increase in signal quality (n=10).
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".