Evaluating the Performance of Machine Learning Techniques for Cancer Detection and Diagnosis

Abstract

Machine Learning (ML) techniques find value in healthcare due to its ability to process huge data sets and convert them into clinical insights. These insights help physicians and healthcare providers in planning, quick decision making and providing timely care to patients with higher accuracy, lower costs and increased customer satisfaction. This paper evaluates the performance of ML algorithms for cancer detection and diagnosis. It compares the performance of different cancer detection algorithms for different types of datasets. It also investigates how the performance of the ML algorithms improve with the use of feature extraction methods, specifically for lung cancer. Developing generic feature extraction methods still remain as a challenge. Healthcare providers need to reorient their treatment approach for diseases like cancer with more focus on detecting them at a very early stage so as to maximize the chances of recovery for the patients.