An analytics framework for inverse cascade classification in cancer risk assessment

Understanding chronic disease risk factors helps people make informed lifestyle choices and live longer; however, patients often struggle to stay healthy because health information can be overwhelming, and diseases have multiple causes. We map survey-derived predictors to specific organ systems using the WHO International Classification of Diseases (ICD) to pinpoint cancer subtypes and find a third-dimensional relationship with mental health. Using our framework, we identify high-impact risk factors, highlight preventive measures, and estimate the likelihood of age-related cancer to guide lifestyle interventions. Using Knowledge Graph-driven feature selection, we also introduce an inverse cascading deep learning (DL) multilabel classifier, a feedforward neural network (NN), for benign majority class filtering and machine learning (ML) classifiers trained solely on minority classes, with minority class-suited hyperparameters and cross-validation. This framework correlates risk factors with the ICD chapters, revealing the interconnected relationship among chronic illness, risk factors, and mental health.