Predicting factors influencing survival of breast cancer patients using logistic regression of machine learning
Huda Kutrani, Saria Eltalhi, Naeima Ashleik · 2021
Breast cancer is the most common cancer in women worldwide. majority of studies focused on identifying factors that have an effect on survival using statistical methods, but few studies have used machine learning techniques. This study aimed to use machine learning to develop a Logistic Regression Model for identifying the factors that influence the survival of patients with breast cancer. Dataset gained from patients’ records at the Department of Oncology (DoO), the Benghazi Medical Centre during a period from 2018 to 2019. The dataset contained 9 predictors and class attributes (survival). The prediction model was built using logistic regression. Our study showed that both the training dataset and the testing dataset have achieved above 94% as a performance of the logistic regression model. The important factors identified in this study were cancer stage classification, tumour size, number of positive lymph nodes, and types of primary treatment. Also, odds ratio results showed that "Mastectomy" as the treatment, Grade II, "positive nodes of 1 to 3", and "tumour size, 2 cm or less” increase survival. Our study concluded that the logistic regression classifier showed high accuracy hence machine learning methods could be used as alternative predictive tools in breast cancer survival studies.