ARCHIVES

Since cancer is one of the main causes of mortality worldwide, risk assessment and early detection are crucial. A machine learning-based method for identifying cancer patients using both clustering and classification techniques is presented in this paper. For analysis, a sizable dataset comprising more than 50,000 patient records with clinical, lifestyle, and demographic characteristics was used. Managing missing values, encoding categorical variables, and getting the dataset ready for modeling were all part of the data preprocessing step. Patients were divided into low, medium, and high risk groups using K-Means clustering. The models' prediction power was significantly improved by using these risk groupings. Several machine learning techniques were developed and assessed, such as Support Vector Machine (SVM), Decision Tree, Random Forest, and XGBoost. The models were evaluated using ROC analysis, confusion matrix, and typical assessment measures like precision, recall, and F1-score. Additionally, a prediction system was created that enables users to enter patient information and obtain probability estimation and cancer risk. The suggested method shows how well clustering and classification techniques can be combined for healthcare prediction and decision assistance.