Abstract

Breast cancer is the most common form of cancer present in women worldwide. It occurs in women of diverse ages and backgrounds with a higher risk of the disease arising in women who are menopausal. The cause of breast cancer is uncertain, and the disease can resurge even after diagnosis and treatment have been carried out. A typical approach used to detect the recurrence of breast cancer is through a mammogram test and although effective, this method can be highly invasive, expensive and studies have expressed the risks of false anomalies being detected during a mammogram scan which could lead to a patient being subjected to a series of unnecessary procedures and tests.

This demands for a more reliable and unintrusive system to be established to handle these shortcomings faced when using the conventional breast screening methods and a smart health system to be able to tackle these issues. The use of machine learning to perform predictions in the health care sector has been emphasized in several studies and with the advancements in technology, and machine’s ability to train and learn supervised or unsupervised with a selected model, the accuracy level of these predictions can be weighed and the model with the best predictive accuracy score used to develop a smart system to aid in predicting breast cancer.

This project is aimed at creating a smart health system that utilizes a suitable machine learning algorithm to predict the likelihood of breast cancer recurring in order to mitigate its associated risks and begin early treatment. A suitable breast cancer dataset was selected from the UCI machine learning repository and the data was pre-processed with exploratory data analysis conducted following the data cleaning, finally, the prepared data was modelled to carry out predictions.

Five machine learning classification models were selected for comparison of the best prediction accuracy score. The machine learning models selected are the Support Vector Machine classifier, KNeighbors classifier, Decision Tree classifier, Naive Bayes classifier and random forest classifier. The random forest classifier model produced the highest prediction accuracy score after the data was trained and tested.

Smart Health Prediction System for Breast Cancer Recurrence