Prediction of Personalized Medicine using Deep Learning

Abstract

Personalized medicine analysis the individual patient’s tumor and determines the combination of drugs that will work for that particular individual patient. To do this task, thousands of genetic mutations in cancer tumor must be distinguished into mutations that contribute to the growth of tumor (drivers) and that do not contribute to the growth of tumor (passengers). Currently the interpretation of genetic mutation is done manually and is time consuming process where a pathologist review and classify every genetic mutation manually based on evidence text based clinical literature. In this paper, we propose a deep learning technique which classifies the genetic variations automatically. The results indicate that deep learning algorithms are more accurate than the baseline machine learning algorithms.