Process Mining of Disease Trajectories in MIMIC-III: A Case Study

A temporal disease trajectory describes the sequence of diseases that a patient has experienced over time. Electronic health records (EHRs) contain coded diagnoses can be mined to find common and unusual trajectories have potential generate clinically valuable insights into relationship between diseases. Disease are typically identified by timestamped diagnostic codes very similar event logs activities used in process mining, we believe models produced using mining tools techniques. We explored this through case study sequences from publicly available MIMIC-III database de-identified EHR data. In paper, present an approach recognised unique nature based on sequenced pairs tested for directionality. To promote reuse, developed set log transformations mine standard tools. Our method was able produce effective relevant MIMIC-III, demonstrates feasibility applying modelling.