Regret theory-based clinical efficacy evaluation method with three-way decision.

Area of Science:

• Medical Informatics
• Decision Science
• Computational Medicine

Background:

• Clinical efficacy evaluation is vital for healthcare quality and patient outcomes.
• Current methods lack interpretability and accuracy, hindering precision medicine.
• Patient response to treatment is a dynamic, temporal, three-way classification problem.

Purpose of the Study:

• To propose an interpretable and accurate method for clinical efficacy evaluation using three-way decision (TWD) theory.
• To address limitations of existing methods in decision-making interpretability and accuracy.
• To develop a data-driven TWD model for temporal, incomplete, and hybrid decision systems.

Main Methods:

• Introduced a multi-granularity unbalanced temporal incomplete hybrid decision system (MGUTIHDS).
• Transformed MGUTIHDS into a normalized system (NMGUTDS) using reference sequences for incomplete data.
• Segmented NMGUTDS by treatment regimen and developed a data-driven TWD model based on loss functions and conditional probabilities.
• Devised a regret theory-based three-way efficacy evaluation method with algorithmic implementation.

Main Results:

• The proposed regret theory-based TWD method demonstrated superior performance.
• The model achieved high accuracy in evaluating clinical efficacy for treatment regimens.
• Application to rheumatoid arthritis data showed the method's effectiveness in regimen recommendation.

Conclusions:

• The developed TWD method offers an interpretable and accurate approach to clinical efficacy evaluation.
• The approach effectively handles temporal, incomplete, and hybrid data in healthcare decision-making.
• This work advances precision medicine by providing a robust framework for treatment regimen analysis and recommendation.