Skin tone corrected pulse oximetry models evaluated through reflective pulsatile Monte Carlo simulations.

Abstract

Significance: Skin tone bias in pulse oximetry is frequently reported to lead to overestimation of arterial blood oxygen saturation ( ${\mathrm{SpO}}_2$ ) in individuals with darker skin tones. We introduce clinically applicable, skin tone corrected pulse oximetry models to address this disparity and promote more equitable patient care.

Aim: We aim to mitigate skin tone bias in pulse oximetry by integrating a melanin correction factor into the conventional ratio-of-ratios approach. We employ pulsatile Monte Carlo simulation to systematically map the "ratio of ratios" ( $R$ ) to ${\mathrm{SaO}}_2$ across varying melanin concentrations, thereby enabling comprehensive quantitative assessments of different ${\mathrm{SpO}}_2$ correction models.

Approach: A six-layered skin model is implemented to approximate the microstructure of human skin, allowing diffuse reflectance computations through Monte Carlo methods. Pulsatile effects are captured by modeling diastolic and systolic phrases, facilitating the calculation of $R$ values under diverse ${\mathrm{SaO}}_2$ and melanin conditions.

Results: Skin tone induced error is quantified and mitigated by incorporating a melanin correction factor into conventional ratio-of-ratios. The inclusion of a melanin-dependent variable results in a substantial reduction in skin tone bias.

Conclusions: Monte Carlo simulations offer precise control over both melanin content and ${\mathrm{SpO}}_2$ , conditions that are challenging to replicate in vivo, and enable the development of robust ${\mathrm{SpO}}_2$ correction models. These findings represent a valuable step toward more equitable pulse oximetry practices, ultimately improving patient care across diverse skin tones.