Breast Cancer Prediction: A Random Forest-based System with Expert Validation

Breast cancer (BC) is a fatal invasive disease among women that impacts women globally. It is listed as a significant disease among Malaysian women. Early detection and accurate diagnosis are important to improve the treatment outcome of a patient, as advanced stages of BC can increase fatality rates. The conventional methods of diagnosis are effective, but they face challenges such as high cost, radiation exposure, and the need for specialized operators. Therefore, this study focuses on developing a BC prediction system using a Random Forest (RF) algorithm. It is trained using the "BC Wisconsin (Diagnostic) Data Set" from Kaggle, consisting of 570 records with eight critical attributes selected for prediction. The algorithm and system are developed using Python and evaluated on accuracy, precision, recall, and F1-score, achieving 91.23%, 90.70%, 86.67%, and 88.89%, respectively. The algorithm was integrated with AdaBoost and XGBoost to add the experimental value, resulting in a better result than a single RF. Expert validation by a specialist confirmed the reliability of the dataset and accuracy of the prediction system, highlighting its potential to be a valuable tool for early BC detection. The study concludes that the RF-based system provides robust predictions, making it a promising approach for enhancing BC diagnostic processes.