Shining Light on Dark Skin: Pulse Oximetry Correction Models

Abstract

Pulse oximeters are medical devices used to assess peripheral arterial oxygen saturation ($SpO_{2}$) noninvasively. In contrast, the “gold standard” requires arterial blood to be drawn to measure the arterial oxygen saturation ($SaO_{2}$). Devices currently on the market measure $SpO_{2}$ with lower accuracy in populations with darker skin tones. Pulse oximetry inaccuracies can yield episodes of hidden hypoxemia (HH), with $SpO_{2} \geq 88\%$, but $SaO_{2}< 88\%$. HH can result in less treatment and increased mortality. Despite being flawed, pulse oximeters remain ubiquitously used; debiasing models could alleviate the downstream repercussions of HH. To our knowledge, this is the first study to propose such models. Experiments were conducted using the MIMIC-IV dataset. The cohort includes patients admitted to the Intensive Care Unit with paired ($SaO_{2}, SpO_{2}$) measurements captured within 10min of each other. We built a XGBoost regression predicting $SaO_{2}$ from $SpO_{2}$, patient demographics, physiological data, and treatment information. We used an asymmetric mean squared error as the loss function to minimize falsely elevated predicted values. The model achieved $R^{2}= 67.6\%$ among Black patients; frequency of HH episodes was partially mitigated. Respiratory function was most predictive of $SaO_{2}$; race-ethnicity was not a top predictor. This single-center study shows that $SpO_{2}$ corrections can be achieved with Machine Learning. In future, model validation will be performed on additional patient cohorts featuring diverse settings.