Artificial Intelligence-Enabled Detection and Assessment of Parkinson's Disease using Nocturnal Breathing Signals

Summary

Landmark study demonstrating that an AI model can detect Parkinson's disease (PD) and track its progression from nocturnal breathing signals captured contactlessly via in-home radio waves. The model uses a transformer with attention over breathing time series to infer PD severity without any wearable device on the patient. The AI achieves AUC 0.90 on held-out and 0.85 on external test sets for PD detection, and R=0.94 for severity estimation aligned with the UPDRS clinical scale.