Using Machine Learning and Natural Language Processing to Detect Cancer
Despite great advances in cancer treatment over the past several decades many forms of cancer especially advanced stage cancer remain resistant to even the strongest doses of radiation and chemotherapy. As a result patient survival often rests largely on early detection via radiological imaging. Screening patients this way has saved countless lives but suffers from inaccuracies even when expert radiologists perform the examinations due to the difficulty of distinguishing between benign and cancerous abnormalities. The goal of this project is to improve the accuracy of early detection mechanisms by building a machine learning model to classify the condition of the patient based on screening images and radiology reports.

I first became interested in machine learning when I took 6.036 last spring. I was amazed by the power of the algorithms we learned so I hoped to explore applications of these algorithms in more depth. Professor Barzilay introduced some of her research during class and this project sounded like the perfect opportunity to explore my interests while developing a tool that could seriously benefit patients with cancer.