Cuckoo Search Optimized Random Forest for Breast Cancer Prognosis

Breast cancer remains a major global health concern due to its high mortality rate, particularly when diagnosis occurs at an advanced stage. Accurate and early differentiation between benign and malignant tumors is therefore critical for improving patient outcomes. Conventional diagnostic practices largely depend on manual assessment and clinical expertise, which may lead to subjective variability in decision-making. To overcome this limitation, this study presents an automated machine learning–based screening framework for breast cancer prognosis. The proposed approach employs a Random Forest classifier for tumor classification, with feature space transformation performed using Principal Component Analysis to reduce dimensionality and enhance discriminative capability. To further improve predictive performance, the hyperparameters of the classifier are optimized using the Cuckoo Search algorithm. The model is trained and assessed using the benchmark breast cancer dataset from the UCI Repository. Experimental results demonstrate that the optimized framework achieves an accuracy of 98% on the test dataset, indicating strong classification capability. The proposed method offers a reliable and efficient computational tool that can assist clinicians in early-stage breast cancer diagnosis.