Hu Submits NIH R01 Application

Kudos to Xiao Hu, professor, and his entire team for the submission of their NIH R01 application entitled "Extremely Scaling up Training Data Size for Learning Accurate Ambulatory Atrial Fibrillation Detection." This proposal requests funding for a five-year period with a start date of September 1.

Atrial fibrillation and atrial flutter (AF, collectively referred here) together are the most common arrhythmia affecting 33.5 million people globally with a growing prevalence. AF is associated with significant morbidity and mortality, including 20% of all strokes, 33% of hospitalizations related to cardiac arrhythmias, and a two-fold increase in risk of death. Long term AF may result in cognitive decline and dementia and precipitate heart failure. However, the most dreaded complication of AF is ischemic stroke. Therefore, diagnosis of AF at scale holds the key to reduce various risks associated with AF. However, this goal is not achieved yet even though a constellation of enabling technologies for AF diagnosis at scale is already available. Modest performance of PPG-based AF algorithm remains a critical impediment to fully realize its potential as a widely available AF-screening tool at scale. Our novel strategy to label a large EHR-linked repository of physiological signals will result in a database of millions of records of AF and non-AF – an unprecedented asset for the research community to fully exploit the power of deep neural networks to develop accurate PPG-based AF algorithm to enable AF diagnosis at scale.

Specifically, we will pursue the following three aims: 1) To develop PPG-based AF detection algorithms based on a large EHR-linked physiological signal repository and validate the algorithms on prospective data from free-living settings; 2) To develop transfer learning approaches to individualize PPG detection algorithms; 3) To develop neural network compression approaches to enable AF detection on edge devices.