Process Mining the Trauma Resuscitation Patient Cohorts

Area of Science:

• Healthcare analytics
• Medical informatics
• Clinical data analysis

Background:

• Analyzing patient cohorts and their associated treatment patterns is crucial for improving medical processes.
• Existing methods may not effectively quantify the associations between patient groups and specific medical interventions.

Purpose of the Study:

• To present a novel framework for analyzing associations between patient cohorts and trauma resuscitation procedures.
• To quantify the relationship between identified patient cohorts and their corresponding treatment patterns.

Main Methods:

• Developed a framework to quantify associations between patient cohorts and treatment patterns.
• Evaluated the framework on a trauma resuscitation dataset from a level 1 trauma center.
• Utilized clustering for patient cohort discovery and process mining for pattern identification.

Main Results:

• The algorithm's learned weights significantly improved patient similarity measurements.
• Identified four distinct patient cohorts through clustering.
• Discovered statistically significant trauma resuscitation patterns using process mining.

Conclusions:

• The developed framework effectively analyzes associations between patient cohorts and medical procedures.
• The approach enhances patient similarity metrics and reveals significant treatment patterns.
• The framework is generalizable for analyzing other medical processes beyond trauma resuscitation.