Machine Learning for Clinical Prediction
The ever-increasing amount of data from medical devices and records sequencing and other sources has enormous potential to change the way patients are treated: to detect diseases earlier and make diagnostics more accurate and interventions more applicable. However the size and complexity of these data make them difficult to analyze. Deep learning is promising as it enables without hand-engineering features learning a powerful representation for discriminative modeling to classify/regress or predict events of interest. This project seeks to help physicians use data-driven approaches to improve patient outcomes. Using the MIMIC III database a computational architecture is developed to predict sepsis and mortality in a real-time forward-facing setting.

I’m studying computer science and molecular biology at MIT with a minor in statistics and data science. It’s very important to me to make an impact with my work which is one of the reasons that I choose biology as my application field. I’m also interested in AI techniques. With this project I hope to learn how to take the algorithms and skills I’ve developed here beyond the classroom and help heal people.