I am a Ph.D. student in the Data-Driven Inference Group at MIT CSAIL, advised by Professor John Guttag. I did my undergraduate in Applied Mathematics at Harvard University. I completed an internship at Google in Summer 2014 and an internship at Grand Rounds in Summer 2017. I am interested in using machine learning to improve healthcare, and the role of public policy in shaping research and education in science and technology.
My research focuses on developing models to predict rare adverse events. A primary challenge in clinical risk model development is the scarcity of data relevant to the patient population and task of interest: there are often few data and a large class imbalance. My work leverages knowledge from auxiliary sources (e.g., data from similar patient populations, expert-encoded ontologies) to improve the performance of these risk models.
Recently, I have been interested in effectively integrating heterogeneous clinical data modalities for predicting patient risk of adverse events such as mortality. I am interested in the relative utility of these modalities in different care scenarios, and I am investigating this question using the publicly available MIMIC dataset. I worked with Tristan Naumann on using existing ontologies of clinical concepts to map information encoded in different EHR versions to a common semantic space to enable better transfer of predictive models across EHR version changes. In addition, I have worked on predicting adverse events in patients after cardiac surgery in collaboration with clinicians at the Massachusetts General Hospital and the Brigham and Women's Hospital. I also collaborated with domain experts at the Massachusetts General Hospital Institute of Health Professions to automatically detect and diagnose speech and language impairments in children.
See my C.V. for a complete list of presentations and publications.