Machine Learning and Genetic Programming to Enhance Diagnosis and Prognosis in Breast Cancer Care

Dissertation presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data Science ; Breast cancer is a pathological condition characterized by uncontrolled cell multiplication within mammary tissue and is undoubtedly the malignant neoplasm that most affects women worldwide. Nevertheless, comprehending its etiology and its therapeutic strategies has progressively evolved throughout history and, in recent years, significant advancements in medicine and artificial intelligence have bolstered the deployment of machine learning and genetic programming algorithms. The use of these algorithms has not only caused a significant impact but has also demonstrated remarkable efficacy in developing predictive models for precise and expeditious patient disease diagnosis and prognosis. The process that guided my research began with the identification and definition of the primary problem: the uncertainty in breast cancer diagnosis and prognosis, along with the limited availability of diverse diagnostic and prognostic algorithms for accurate results. This challenge served as the impetus for my thesis. Consequently, my research pursued the objective of applying machine learning and genetic programming techniques to improve the accuracy and efficiency of breast cancer diagnosis and prognosis. In terms of diagnosis, the primary problem to address is distinguishing between malignant and benign breast cytology. Regarding prognosis, my research focused on two main problems: firstly, distinguishing between recurrent and non-recurrent breast cancer, and secondly, predicting the recurrence time for recurrent cases as well as the disease-free time for non-recurrent cases. The research encompasses analyzing data from the Breast Cancer Wisconsin repository. This includes understanding the data, preprocessing, feature selection, and implementing and training machine learning and genetic programming algorithms to accurately diagnose and prognose cancer. Subsequently, the ...