A novel smartphone-based diagnostic app using multimodal digital biomarkers demonstrated potential to achieve >80% sensitivity and specificity for detecting OSA compared to standard home sleep tests.
Cross-Sectional (n=456)
Sí
Does a smartphone-based diagnostic app accurately predict obstructive sleep apnea in patients referred for suspected OSA?
A novel smartphone-based diagnostic app using multimodal digital biomarkers demonstrates promising accuracy (>80% sensitivity and specificity) for detecting obstructive sleep apnea compared to standard home sleep tests.
Abstract Introduction Early detection and treatment of obstructive sleep apnea (OSA) improves health outcomes and reduces healthcare costs, yet under-recognition and limited access to diagnostic testing remain major barriers. Conventional in-lab and home sleep studies require specialized sensors and personnel, creating significant delays and costs. There is a strong need for accessible, low-cost diagnostic tools that can identify OSA beyond traditional symptom-based screening. We developed novel smartphone-based application that integrates multiple data modalities which show promise of detecting OSA; including facial imaging, video assessments (speech, swallowing, blinking), heart rate variability, and targeted questionnaires. A pilot study is underway to assess the utility of the app for predicting OSA risk. Methods In this pilot study, participants referred for suspected OSA across multiple sleep centres completed the app-based assessment prior to undergoing a standard home sleep test (HST). Associations between app digital biomarkers and HST derived Apnea-hypopnea index (AHI) and oxygen desaturation index (ODI) were analysed using correlation and multivariable regression to assess predictive validity. Results To date, 456 participants have been analysed (37.5% female; age distribution: 3% aged 20, 28% aged 20-40, 46% aged 40-60, 15% aged 60-70, and 7% aged 70-99). Among participants with available AHI data (n = 339): 17% had AHI 5, 27% had AHI 5-14, 24% had AHI 15-30, and 31% had AHI 30. Racial distribution included 59% White, 23% Asian, 6% Black or African American, 4% Middle Eastern or North African, and 8% other. Preliminary modelling indicates the potential to achieve 80% sensitivity and specificity for OSA detection relative to AHI ≥15 defined diagnosis. Conclusion This novel smartphone-based diagnostic app demonstrates promising accuracy in identifying OSA using multimodal digital biomarkers. Early results support its potential as a home-based, low-cost diagnostic solution. A larger prospective validation study across diverse populations is warranted to confirm clinical performance and generalizability. This abstract is funded by: Resmed
Wimms et al. (Fri,) conducted a cross-sectional in Obstructive Sleep Apnea (OSA) (n=456). Smartphone-based diagnostic app vs. Standard home sleep test (HST) was evaluated on OSA detection relative to AHI ≥15 defined diagnosis. A novel smartphone-based diagnostic app using multimodal digital biomarkers demonstrated potential to achieve >80% sensitivity and specificity for detecting OSA compared to standard home sleep tests.