Key result
The proposed physics-driven r-PPG algorithm achieved an overall mean absolute error of 4.17 bpm across diverse skin tones, outperforming existing facial aggregation (4.49 bpm) and SNR weighting (4.81 bpm) methods.
Why the study?
Existing computer vision methods that estimate heart rate from facial videos exhibit biased performance against dark skin tones, limiting contactless sensing for telemedicine.
Does a novel physics-driven r-PPG algorithm improve heart rate estimation accuracy in subjects with diverse skin tones compared to existing benchmark methods?
Cross-Sectional (n=54)
No
Does a novel physics-driven r-PPG algorithm improve heart rate estimation accuracy in subjects with diverse skin tones compared to existing benchmark methods?
Absolute Event Rate: 4.17% vs 4.49%
A novel physics-driven r-PPG algorithm improves contactless heart rate estimation accuracy and reduces performance bias across diverse skin tones and recording conditions, facilitating more inclusive telemedicine applications.
May improve remote HR monitoring equity across skin tones; leaves open prospective clinical validation.
Heart rate (HR) is an essential clinical measure for the assessment of cardiorespiratory instability. Since communities of color are disproportionately affected by both COVID-19 and cardiovascular disease, there is a pressing need to deploy contactless HR sensing solutions for high-quality telemedicine evaluations. Existing computer vision methods that estimate HR from facial videos exhibit biased performance against dark skin tones. We present a novel physics-driven algorithm that boosts performance on darker skin tones in our reported data. We assess the performance of our method through the creation of the first telemedicine-focused remote vital signs dataset, the VITAL dataset. 432 videos (~864 minutes) of 54 subjects with diverse skin tones are recorded under realistic scene conditions with corresponding vital sign data. Our method reduces errors due to lighting changes, shadows, and specular highlights and imparts unbiased performance gains across skin tones, setting the stage for making medically inclusive non-contact HR sensing technologies a viable reality for patients of all skin tones.
No takes yet. Share an insight, caveat, or question.
Chari et al. (2020) reported a cross-sectional. Physics-driven r-PPG algorithm vs. Facial aggregation and SNR weighting r-PPG methods was evaluated on Mean Absolute Error (MAE) of heart rate estimation. The proposed physics-driven r-PPG algorithm achieved an overall mean absolute error of 4.17 bpm across diverse skin tones, outperforming existing facial aggregation (4.49 bpm) and SNR weighting (4.81 bpm) methods.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: