Russian AI Assistants Aim to Detect Causes of Poor Sleep
AI prototypes can analyze sleep structure and identify breathing disorders simultaneously
Researchers at the Research Institute of Neurosciences at Samara State Medical University have developed two AI-based software prototypes designed to diagnose sleep disorders.
The first is a software module that automatically detects breathing disorders from audio recordings using mel-spectrogram analysis. A mel-spectrogram converts breathing and snoring sounds into visual representations that artificial intelligence algorithms can process.
The second prototype is a program that automatically interprets polysomnography data and classifies sleep stages based on electroencephalogram readings. It relies on a deep learning model that processes EEG signals and assigns each 30-second segment of a recording to a corresponding sleep stage.
Scaling Sleep Monitoring
The developers tested the functional prototype on real-world recordings and received official registration for the software. They now plan to conduct clinical trials and integrate the system into medical information platforms.
