Abstract 4143416: Optical Absorption Figure of Merit for Photoplethysmography-Based Cuffless Monitoring of Blood Pressure Varies Across a Diversity of Skin Tones in the Fitzpatrick Scale

Introduction: Well-publicized studies have shown that photoplethysmography (PPG), a key technology for pulse oximeters and consumer wearables, may have limitations for users with darker skin tones. This study assessed the correlation between skin tone and figures of merit for PPG sensing in a multiwavelength, reflection-mode, fingertip-worn PPG device. The device measures optical signals on the underside of the hand, where melanin density is relatively low; thus, this study hypothesized that the figures of merit would show an insignificant correlation with skin tone. Methods and Results: A study at Houston Cardiology Consultants saw trained professionals tend to 174 participants (Table 1). Subjects wore a multiwavelength, reflection-mode, PPG-based fingertip sensor. Subject metadata, including skin tone, were collected manually. PPG datasets were analyzed by Valencell, Inc. using the company’s standard figures of merit: AC/DC ratio, Signal-to-Noise-And-Distortion (SINAD), and Relative Peak Ratio (RPR). SINAD and RPR represent motion-artifact resistance and hemodynamic consistency, respectively. The AC/DC ratio measures the strength of the pulsatile PPG waveform against unwanted optical scatter due to melanin and other skin chromophores. Using the Fitzpatrick Scale to quantify skin tone (Figure 1), its correlation with these figures of merit was assessed. No correlation was found between the Fitzpatrick value and SINAD or RPR across optical wavelengths. However, a strong correlation between the Fitzpatrick value and AC/DC ratio was observed for all wavelengths, with the green AC/DC ratio showing the highest correlation (Table 2). Conclusion: The observed disparity in AC/DC ratio with skin tone must be considered when developing PPG-based physiological monitors to ensure they operate effectively across all Fitzpatrick values, even when these monitors target skin regions with relatively low melanin density. Engineering approaches that rely solely on green wavelengths of light, the most commonly used in consumer wearables, will face greater challenges than those utilizing longer wavelengths.