Process Mining for Healthcare Process Improvement

Project Team: William K. Cheung (In-charge), Jiming Liu, Byron Choi, Joseph Ng

External Collaboration: Benjamin Yen (Business, HKU)

Background and Research Problems:

Healthcare process efficiency enhancement aims to make best use of the available resources to improve the throughout rate of the healthcare system. The resulting shortening on patient waiting time normally ends up with patient experience which is especially desirable for health critical patients who require timely treatment to increase the chance of recovery.

This project is to develop robust process mining (also called workflow mining) algorithms which can identify patterns in real workflow logs, which in turn can support the investigation of strategies to enhance healthcare process efficiency. It covers the use of stochastic models for i) bottleneck detection in temporal social networks of stakeholders in hospitals, and ii) control flow discovery to uncover the healthcare workflows in practice.

Regarding research issues, for the first part of the project, we will build upon our preliminary work on stochastic network motifs detection and extend it to some specifically constrained stochastic temporal network motifs to model workflow bottlenecks causing patients’ long waiting time.

For the second part of the project, we will develop stochastic petri net related models for robust recovery of control flow structure in workflow log data with a high level of noise. As control flow in a hospital can be complicated, careful control on the recovery results on the granularity of the discovery results will be needed. The study will first be evaluated on synthetic data and later on hopefully real data from hospitals.