Methods of Machine Learning in Oncology

Abstract

Malignant diseases are among the leading causes of mortality. The registration of malignant diseases in Bulgaria is mandatory. The information from the National Cancer Registry about the registered patients, their psychosocial distress and the medical establishments provides an opportunity to conduct research with machine learning methods on a large volume of data. The study aims to apply three machine learning algorithms - CHART, Boosting and Bagging to create classification models for determining the rating of a medical institution and the level of distress of a patient. We present results obtained in a comparative study of different classification methods.