On mining clinical pathway patterns from medical behaviors.

Area of Science:

• Medical Informatics
• Process Mining
• Clinical Pathway Analysis

Background:

• Clinical pathway analysis is crucial for standardized medical procedures.
• Clinical pathway pattern mining identifies essential medical behaviors and their temporal relationships.
• Existing methods lack precise quantification of temporal orders for critical medical behaviors.

Purpose of the Study:

• To develop a new process mining approach for clinical pathway pattern discovery.
• To identify critical medical behaviors and their quantified temporal orders within clinical pathways.
• To address limitations in existing clinical pathway pattern mining techniques.

Main Methods:

• Utilized process mining to analyze clinical pathway workflow logs.
• Developed a novel approach to discover clinical pathway patterns based on a minimum support threshold.
• The approach identifies critical behaviors, their order, and quantified temporal relationships.

Main Results:

• Evaluated on real-world datasets from six diseases at Zhejiang Huzhou Central hospital.
• The proposed approach demonstrated reduced processing time compared to general sequence mining.
• Generated fewer clinical pathway patterns with linear scalability against increasing data size.

Conclusions:

• The developed approach is applicable for discovering clinically relevant pathway patterns.
• It effectively covers frequent and regularly encountered medical behaviors in clinical practice.
• Shows significant promise for advancing clinical pathway analysis research.