Exploring Object Centric Process Mining with MIMIC IV: Unlocking Insights in Healthcare

The vast Medical Information Mart for Intensive Care (MIMIC IV) dataset offers a goldmine for process mining in healthcare, yet traditional approaches often struggle with complex object interactions like patients, medications, and diagnoses. Object Centric Process Mining (OCPM) unlocks deeper insights into patient care, leading to improved coordination, resource allocation, and ultimately, better patient outcomes. This work explores how OCPM overcomes these limitations. By analyzing data flow within the MIMIC IV dataset through an OCPM lens, we illuminate intricate relationships between objects across clinical processes like heart patient or medication administration. By leveraging OCPM on MIMIC IV data, this study offers a novel perspective on emergency department processes of healthcare. We propose a method that leverages the inherent structure of MIMIC IV to directly extract relevant objects, visualizing the convergence and divergence picture and their relationships, bypassing the traditional conversion from XES to OCEL. Our approach focuses on key clinical entities like patients, hospitals admission, and transfer of patients across various departments in hospitals to construct an Object Centric Event Log (OCEL) that captures patient journeys within the hospital system. This direct object-centric approach aims to streamline the process discovery phase and potentially unlock new insights into patient flow patterns and clinical care pathways.