AI and Hierarchical clustering techniques for accurate patient stratification

Abstract Graph-based methods for data representation and analysis are well suited for encoding both data points and their interrelationships. This approach integrates data and topology, enabling the representation of interrelated information. In this study, we represent patient cohorts as cohort graphs and discuss their application for real-world patient data. We particularly focus on developing methods to cluster patients with similar symptoms and examine how bias parameters (such as sex and age group) influence interlinking within CGs, thereby improving results for accurate patient stratification and personalized decision-making in a clinical context. In particular we illustrate how by considering sex and age groups we can improve the symptom-clustering of a patient population with lung and gastro-intestinal cancer. Finally, we discuss the essential role of high-performance computing (HPC) in upscaling analytical methods for CGs.