Can Oximetry-Derived Parameters Predict 3-Year Cardiovascular Mortality?

As obstructive sleep apnoea (OSA) is associated with cardiovascular disease (CVD), we are keen to investigate whether future CVD outcomes can be predicted by measuring OSA-associated hypoxemia. The apnoea-hypopnoea index (AHI), the gold standard for OSA diagnosis, has been shown to be inadequate for predicting CVD outcomes, and studies have explored oximetry-derived parameters as potential predictors. While many studies have proposed individual oximetry-derived parameters as promising predictors, the combined predictive performance has not been thoroughly investigated. This study aims to evaluate the combined predictive performance of demographic information, lifestyle habits, and oximetry-derived features for 3-year CVD mortality. The results demonstrate that oximetry-derived parameters alone are insufficient for prediction (F1 scores: 58.1% and 53.65% in the training and testing datasets, respectively), but perform well when combined with demographic and lifestyle features (F1 scores (p-values <0.05): 79% and 77.47% in the training and testing datasets, respectively). These findings highlight the need to incorporate additional covariates in future studies and suggest the possibility developing a model to predict future CVD events or clinical use.