Predicting recurrence of breast cancer to select the best treatment option

Abstract

"

Breast cancer recurrence is when the treated cancer reappears in the patient. This second cancer is always stronger and aggressive than the initial cancer. Starting treatment as soon as possible is the way to prevent or delay recurrence. Treatments available for cancer can bring out side effects which might lead to different defects. As a solution author is proposing an AI tool to predict the probability of breast cancer recurrence. This is to be predicted with logistic regression and decision tree. Main features which impact breast cancer recurrence is found through decision tree. These features will be run through a logistic regression model to find the recurrence probability. A public dataset from UCL Machine Learning repository will be used to find the features with highest impact and to train the data model. The trained data model presented a 79% accuracy and 71% testing accuracy. This model was implemented to Flask web framework where the user interfaces are to be made with HTML. The proposed and implemented system was then evaluated with domain experts (Oncologist/Doctors) and Technical Experts to get feedback and validation of the functionalities. "