Comparing Care Pathways Between COVID-19 Pandemic Waves Using Electronic Health Records: A Process Mining Case Study.

Area of Science:

• Health Informatics
• Process Mining
• Epidemiology

Background:

• The COVID-19 pandemic significantly altered healthcare workflows, yet its impact on multidisciplinary care remains under-researched.
• Limited evidence exists on how pandemic-induced changes affected complex care settings.

Purpose of the Study:

• To apply process mining to electronic health record data for analyzing changes in multidisciplinary care patterns during the COVID-19 pandemic.
• To quantify variations in care complexity and identify shifts in healthcare practices across pandemic waves.

Main Methods:

• Utilized timestamped electronic health records from Scottish hospitals for adult COVID-19 patients.
• Employed process mining with the Inductive Miner infrequent (IMi) algorithm and conformance checking.
• Analyzed provider- and activity-level data using Petri Nets, cross-log conformance checking, and graph edit distance (GED).

Main Results:

• The IMi model demonstrated good log fitness and generalization but limited precision.
• Wave 1 care procedures were more structured compared to Wave 2, which showed reduced multidisciplinary engagement.
• Significant differences in care activities were observed between patients with extended versus shorter hospital stays.

Conclusions:

• Process mining effectively reveals differential care complexity in COVID-19 patients and quantifies shifts in healthcare practices.
• Findings highlight changes in care structure and multidisciplinary involvement across pandemic waves.
• Future studies can leverage process mining for operational adherence and understanding service changes during high-pressure periods.