Or Yaacov, MD

The availability of large datasets containing paired genomes and clinical data has been increasing in the past 15 years, through large collaborative consortia, private companies, and biobanks. However, studies based on these datasets have struggled to provide meaningful translational insight for various common diseases. This is in part due to the complex genetics and regulatory mechanisms of these diseases, and in part due to noisy or poorly defined clinical phenotypic data.

My training is in clinical medicine, human genetics and computer science and my research work combines all these three disciplines. My overarching research goal is to understand the functional roles of coding (genes) and non-coding (regulatory elements) genomic features and how they affect human phenotypes. My current and previous work aims to provide accurate and scalable computational methods for the discovery of regulatory elements affecting gene expression, diseases, and other clinical phenotypes. My approach involves using machine learning and AI to augment large clinical datasets with specific tissue- and cell-level data that cannot be easily obtained from living participants. As program lead for deciphEHR, a genomic medicine initiative at NYULH and the Center for Human Genetics & Genomics (directed by Aravinda Chakravarti, PhD), I utilize novel AI methods to extract nuanced clinical data from the full scope of electronic health records (EHR), which current datasets lack as they often contain crude summary codes.