Predicting cardiovascular risk from sleep patterns using deep learning
Description: Sleep and cardiovascular system are intrinsically related and synchronized by circadian rhythm. Considerable effort has been made to investigate the associations between sleep disorders and various cardiovascular diseases (CVDs). However, current risk markers for CVDs are mainly derived from lipid molecules, inflammation, clinical factors, genomic profiles or imaging, without nighttime characteristics of sleep. We hypothesize that polysomnographic recordings during sleep contain unique rhythmic features for predicting cardiovascular risk, independent of other risk markers. We propose to model the interactions between sleep events and CVDs, and create novel polysomnogram-based digital biomarkers by deep learning. These biomarkers are expected to provide non-redundant information, improving the personalized prognosis and monitoring of CVDs.
Funder: American Heart Association
Amount: $100,000
Funding type: research
Funder scheme: Institute - AI and ML Training Grants
Start date: 2019-07-01
End date: 2021-06-30
Investigator: Hongyang Li (lead)
Institutions: University of Michigan, The Regents of the University of Michigan
Topic: Sleep and related disorders
Subfield: Experimental and Cognitive Psychology
Field: Psychology
Domain: Social Sciences
Funded outputs count: 7
Funder award ID: 19AMTG34850176
OpenAlex ID: G2415660164
