Uncovering Complex Relations in Patient Pathways based on Statistics: the Impact of Clinical Actions

Abstract

Process mining is a family of techniques that can aid healthcare organizations in improving their processes. Most existing process mining techniques do not provide insights into the impact that activities can have on the process. Some novel techniques try to address this issue, but these techniques are either not generic in their approach or cannot provide insights into complex relations in organizational processes. We propose a novel and generic approach with the goal of producing insights into statistical relations within healthcare processes. We apply the approach on a public data set on sepsis in an emergency room. We find that the hospital might optimize its process in two respects: (1) their cost-benefit balance for patient care by considering their activities in terms of continuous monitoring and substance administration, and (2) their policies on discharging patients as to ensure patients are not discharged too early and return to the emergency room.