Evaluation of Sleep Apnea in Stroke Patients Using a Portable Smartphone-Based System

Sleep apnea is common but often underdiagnosed after stroke, hindering rehabilitation and functional recovery. Access to sleep studies remains limited for stroke patients, partly due to the inconvenience of current diagnostic tools. The widespread availability and built-in sensors of smartphones make them powerful solutions for mobile health applications, including sleep monitoring. This study aims to detect and evaluate sleep apnea in post-stroke patients using a smartphone and extract multimodal digital biomarkers to characterize their sleep patterns. For that, overnight sleep tests were conducted on 30 subacute stroke patients and 30 age- and sex-matched control subjects, using a smartphone-based system that recorded audio, accelerometer, and oximetry data. Signals were analyzed with custom-made algorithms to compute respiratory, oxygenation, and sleep position indices. Results showed that the apnea-hypopnea index (AHI) was significantly higher in the stroke than the control group (28 $\pm $ 19 vs 14 $\pm $ 11, p=0.006). Moderate-to-severe sleep apnea (AHI $\ge$ 15) was present in 67% of stroke patients and 40% of control subjects. Stroke patients spent more time mouth breathing (22% vs 12%, p<0.001) and sleeping in supine position (67% vs 35%, p<0.001) than controls, contributing to upper airway obstruction. These findings obtained with novel multimodal digital biomarkers improve the understanding of post-stroke sleep apnea patterns and show the potential of smartphones as portable monitoring tools. This approach can facilitate early sleep apnea detection after stroke, thereby reducing its substantial burden and improving rehabilitation outcomes.