Survival Analysis of Breast Cancer Patients in Texas Using Classical and Machine Learning Methods

Background Breast cancer is considered one of the most common cancers in women worldwide. In this study, we used an 11-year cohort of malignant breast cancer survival data in Texas to investigate the factors that might explain why some breast cancer patients live longer than others. Methods We performed standard survival analyses, including generating Kaplan‒Meier survival curves, using the log-rank test, and applying Cox proportional hazards regression to identify the unique features of breast cancer patients and determine the main factors influencing long-term survival. We also conducted a Random Survival Forest analysis for classification and prediction. Finally, we used Mahalanobis matching distance to estimate the average extended lifetime of patients in different groups. Results Our study found that stage, laterality, age, grade, subtypes, progesterone receptor status, the primary site of cancer, patient race, median household income, chemotherapy, systemic therapy, and surgery are significant factors, with stage being the most critical for predicting survival in breast cancer patients. Compared to individuals with localized-stage breast cancer, those with distant-stage cancer had a higher hazard ratio (15.869). We also observed that patients diagnosed at the localized stage had an average survival of approximately 67.34 months, compared to those diagnosed at the distant stage. Conclusion Policymakers should focus on promoting early diagnosis and screening, while also providing additional support to the elderly and those in disadvantaged circumstances.