Evaluating the Bias in Hospital Data: Automatic Preprocessing of Patient Pathways Algorithm Development and...

Area of Science:

• Healthcare Management
• Process Mining
• Simulation Modeling

Background:

• Optimizing patient care pathways is critical for hospitals facing resource scarcity.
• Logistical constraints, such as bed availability, can delay treatments and impact patient flow.
• Unscheduled patient admissions pose unique challenges to maintaining planned hospitalization schedules.

Purpose of the Study:

• To develop a framework for automatically detecting activities in patient pathways influenced by logistical limitations rather than patient needs.
• To differentiate between medically necessary and logistically induced pathway deviations.

Main Methods:

• A novel method transforms historical patient pathway data into labeled and corrected databases.
• Activities are labeled as relevant (patient needs) or irrelevant (logistical constraints).
• Process mining and discrete event simulation were used to quantify the method's impact.

Main Results:

• The algorithm achieved 87% accuracy in identifying logistical influences on unscheduled patient pathways.
• Processed data resulted in an average 40% reduction in pathway variants.
• Simulations indicated significant differences in required bed capacity between raw and processed data.

Conclusions:

• Patient pathway data are a composite of medical needs and logistical realities.
• Identifying and correcting for logistical limitations is essential for unbiased analysis of healthcare data.
• The proposed approach offers a generalizable method for improving patient pathway analysis in hospitals.