An adaptive and interpretable case grouping framework for healthcare payment systems.

Area of Science:

• Health economics
• Medical informatics
• Operations research

Background:

• The diagnosis-intervention packet system in China aims for precise medical service management.
• Current auxiliary catalog formulation lacks systematic methodology, hindering effective case grouping.

Purpose of the Study:

• To develop a systematic case grouping framework for the diagnosis-intervention packet system.
• To enhance the auxiliary catalog's foundation for improved medical insurance reimbursement.

Main Methods:

• A mathematical optimization model minimizing intragroup variation.
• Quantifying case severity using nonnegative LASSO regression and natural breaks.
• An adaptive learning differential evolution algorithm with Q-learning and hypercube learning for self-adaptive, interpretable solutions.
• Tree structures for representing case grouping pathways.

Main Results:

• The framework significantly improves intragroup homogeneity and intergroup heterogeneity of case groups.
• Empirical analysis confirms the framework's effectiveness under established constraints.
• Generalizability demonstrated by successful application to subdividing diagnosis-related groups.

Conclusions:

• The proposed framework provides a theoretically sound and practically feasible method for case grouping.
• Offers flexibility and scalability for developing accurate and equitable medical insurance reimbursement schemes.