Applying Machine Learning to Breast Cancer Gene Expression Data to Predict Survival Likelihood

Abstract

Analyzing the expression level of thousands of genes will provide additional information beneficial in improving cancer therapy or synthesizing a new drug. In this project, the expression of 48807 genes from primary human breast tumors cells was analyzed. Humans cannot make sense of such a large volume of gene expression data from each person. Therefore, we used Machine Learning as an automated system that can learn from the data and be able to predict results from the data. This project presents the use of Machine Learning to predict the likelihood of survival in breast cancer patients using gene expression profiling. Machine Learning techniques, such as Logistic Regression, Support Vector Machines, Random Forest, and different Feature Selection techniques were used to find essential genes that lead to breast cancer or help a patient to live longer. This project describes the evaluation of different Machine Learning algorithms to classify breast cancer tumors into two groups of high and low survival.