Interpreting Breast Cancer Recurrence Prediction Models: Exploring Feature Importance with Explainable AI

Breast cancer, the second deadliest cancer affecting women, remains one of the most prevalent malignancies worldwide, affecting millions of individuals annually. Despite advancements in treatment modalities, recurrence remains a significant concern, posing challenges for patient management and long-term outcomes. This paper aims to provide a comprehensive review of the factors involved in breast cancer recurrence, encompassing patient-related variables, tumour characteristics, and treatment interventions. Patient-related factors such as age, genetic predisposition, lifestyle factors, and comorbidities play crucial roles in determining recurrence risk. Additionally, tumour biology, including hormone receptor status, HER2 expression, and molecular subtypes, significantly influence recurrence patterns. Furthermore, treatment modalities, such as surgery, chemotherapy, radiation therapy, and targeted therapies, impact recurrence rates and patterns.