Deep Learning-Enhanced Association Rule Mining for Interpretable Clinical Decision Support
Abstract
The integration of Deep Learning with Association Rule Mining (ARM) is revolutionizing medical data analysis by uncovering complex, nonlinear patterns within high-dimensional clinical datasets. Traditional ARM techniques, though effective for pattern discovery, face scalability and interpretability challenges. Deep learning models such as CNNs, RNNs, and Autoencoders enhance ARM by enabling robust feature extraction and adaptive rule generation. This chapter explores hybrid DL-ARM frameworks for disease prediction, prognosis, and personalized treatment using EHR, imaging, and genomic data. Case studies and performance analyses highlight significant improvements in accuracy, interpretability, and computational efficiency. The study also addresses challenges related to explainability, privacy, and scalability, proposing future directions for federated and quantum-ready ARM systems.