P060 Explainable machine learning reveals associations between sleep breathing disorder biomarkers and incident type 2 diabetes

Abstract Introduction Obstructive sleep apnoea (OSA) and type 2 diabetes (T2D) show bidirectional relationships that negatively impact overall health. Although this relationship has been extensively studied with multiple reported OSA severity markers derived from polysomnography, no studies have systematically searched for novel OSA biomarkers of T2D. This project sought to investigate if state-of-the-art explainable machine learning (ML) models could identify new OSA biomarkers predictive of incident T2D. Methods We applied explainable ML models to longitudinal MAILES study data from 536 males who were free of T2D at baseline and identified 52 cases of T2D at follow-up (mean 8.3yrs, range 41.9-126.5mths). Beyond ranking biomarker importance, we explored how the explainable ML model approach can identify novel relationships, assist in hypothesis testing, and provide insights into risk factors. Results The top most predictive biomarkers included waist circumference, and novel biomarkers: number of desaturations in non-supine sleep, mean heart rate in supine sleep, and mean hypopnea duration. Explainable ML also identified a strong association between the number of desaturation events and incident T2D (Odds ratio=2.4 [95% CI 1.2-4.8], P=0.013). Finally, the explainable ML model revealed an individualized breakdown of risk factors, opening promising new avenues for more personalized precision sleep medicine. Conclusions Explainable ML supports the role of established biomarkers and reveals novel biomarkers of T2D likely to help guide further hypothesis testing and validation of more robust and clinically useful biomarkers. Although further validation is needed, these proof-of-concept data support the benefits of explainable ML in prospective data analysis.