Research on breast cancer prediction method based on machine learning

Abstract

Every year, the number of deaths increases extremely due to breast cancer. In 2020, the World Health Organization reported that 25% of women were diagnosed with breast cancer. In Peru, breast cancer was the main cause of death from cancer, where 5 women die from this disease every day. Early detection of breast cancer is facilitated by computerized detection and diagnostic (CAD) technologies, which can help people live longer lives. This article seeks to analyze multiple machine learning models to identify breast cancer. For this, we work with two Wisconsin databases donated by Dr. William H. Wolberg, the first set of databases is from 1995 and the second is from 1992, both reports are based on their clinical cases. The main objective of this work is to take advantage of the latest developments in CAD systems and related methodologies for predictions, Decision Tree models, K-Nearest Neighbors (K-NN), Naive Bayes (NB), Logistic Regression, Support Vector Machine, Multi-layer Perceptron and Random Forest were used. When the results are compared, it is found that for the database of the year 1995, the Multi-layer Perceptron algorithm offers the best results, achieving an accuracy of 97.20% and for the second database of 1992, the algorithm that best predicts is the Support Vector Machine with an accuracy of 97.20%.