Analysis and prediction of male breast cancer using non-clinical attributes and comparison to female breast cancer using PLCO dataset

Abstract

This study aims to address this knowledge gap by exploring the epidemiology of MBC and then comparing it to FBC, highlighting any differences between the two and identifying potential risk factors. The available literature contains only limited examples of comparative statistical analysis between breast cancer in males and females, performed using the Prostate, Lung, Colorectal and Ovarian (PLCO dataset. The literature review also revealed a lack of studies that compare the accuracy of prediction of MBC and FBC utilising machine learning algorithms, especially using the same attributes.

Following analysis on PLCO datasets for males and females was conducted: a). statistical analysis and b). classification and prediction using machine learning classifiers. Through the statistical analysis, both male and female breast cancer data were examined to identify potential relationships between confirmed cancer data and independent variables such as smoking, age level, and others. The study utilised 73854 records across 27 attributes for MBC and 76113 records across 24 attributes for FBC. Five significant findings resulted from the statistical analysis that highlight differences in the epidemiology of male and female breast cancer with the most significant being that male participants diagnosed with breast cancer exhibited evidence of an unhealthy prostate, which may be a contributing factor to the development of MBC. It was also found that MBC tends to be diagnosed at a higher age with a more advanced stage compared to FBC.

The results of this research were benchmarked against other research conducted using the PLCO MBC dataset; however, the only study found was that by Li and Mani published in 2021 whose study was limited to MBC. A comparison highlighted significant differences in accuracy as well as ROC value, i.e., Li and Mani’s study achieved accuracy between 0.4 to 0.9985 with an ROC range of 0.39 to 0.67 compared to an accuracy of 99.932% to 99.964% and an ROC range of 0.4360 to 0.7642 in this research. Thus, this research reports a clear gender-based distinction in the factors contributing to the identification of the presence of male and female breast cancer.

Date of Award	2023
Original language	English
Awarding Institution	Western Sydney University
Supervisor	Simi Bajaj (Supervisor)