Stanford researchers have developed SleepFM, an AI model that predicts over 100 diseases using one night of sleep study data.
Key Details
- 1SleepFM was trained on 585,000 hours of polysomnography data from 65,000 individuals.
- 2The AI model can predict 130 diseases from sleep data, including Parkinson's (C-index 0.89) and heart attack (C-index 0.81).
- 3Polysomnography collects brain, heart, respiratory, muscle, and eye movement data overnight.
- 4SleepFM outperformed standard models in classifying sleep stages and diagnosing sleep apnea.
- 5Stanford linked up to 25 years of health records to sleep data for longitudinal prediction studies.
- 6The research is published in Nature Medicine and received NIH and Chan-Zuckerberg Biohub funding.
Why It Matters
AI models like SleepFM show the underexploited potential of physiological imaging data for predicting diverse health outcomes. Their ability to integrate multimodal signals for disease risk stratification could reshape early detection in radiology, sleep medicine, and preventive healthcare.
Source
EurekAlert
Related News
•EurekAlert
AI-Powered OCT Enables Rapid 'Optical Biopsy' for Early Endometrial Cancer Detection
A team at Washington University has developed a catheter-based 3D OCT system with AI to quickly and noninvasively detect early endometrial cancers.
•EurekAlert
AI Clinical Reasoning in Diagnostics and Digital Fatigue in Healthcare
Recent JMIR features explore large language models in clinical diagnostics and digital fatigue among healthcare professionals.
•EurekAlert
KAIST, MIT, Microsoft Develop Efficient AI Image Upsampling for Robotics
KAIST, MIT, and Microsoft have created 'Upsample Anything,' a training-free AI method to restore high-resolution visual data from compressed images with up to 16x improved GPU memory efficiency.