Application of Classification Techniques on Breast Cancer Prognosis

Abstract

The most prevalent form of cancer for females is breast cancer in Americans, and additionally, it is Asia's and the United States' second most common cause of death among females. In the United States in 2009, 40,600 persons died from breast cancer, 400 of whom were men. Clinical breast exams, radiographs, and ultrasounds are all excellent methods for testing for breast cancer today. A strategy for presenting a set of input-output sets to a network is referred to as supervised learning. The subsequent model parameters are updated iteratively to minimize the discrepancy around system prediction and real outcomes for training data. Three classification methods were tested on a breast cancer database: Probabilistic Learning, Logistic Regression, and Neural Net. Experiments demonstrated that Neuro Net categorization surpasses Tree Based categorization and Naïve Bayesian classification in terms of accuracy and precision for breast cancer early detection. Although it is established that the use of Ml techniques can enhance our knowledge of cancer progression, these methods must be confirmed before they're able to be employed in clinical practice. In this paper, we give a review of contemporary ML techniques used in cancer progression modelling. The prediction models talked about here have been trained using an assortment of supervised algorithms for machine learning, as well as varied input features and data. We have put together an inventory of the most recent articles that use such methods for modelling cancer risk or patient outcomes in context with the increasing desire to employ ML methods in cancer studies.