A sequential decision-theoretic model for medical diagnostic system.

Area of Science:

• Artificial Intelligence
• Medical Informatics
• Decision Analysis

Background:

• Traditional diagnostic expert systems can replicate expert errors.
• Decision-Theoretic Models (DTMs) mitigate reasoning errors under uncertainty.
• Sequential diagnosis requires effective uncertainty management and test prioritization.

Purpose of the Study:

• To develop a sequential diagnostic Decision-Theoretic Model (DTM) using Bayesian Networks.
• To address uncertainty in expert systems for medical diagnosis.
• To provide a method for prioritizing diagnostic tests.

Main Methods:

• A sequential diagnostic model based on Bayesian Networks was developed.
• Features were categorized into disease and test features.
• An arithmetic for test priors and feature weight adjustments was devised.
• Bayesian Networks were used for uncertainty representation and propagation.

Main Results:

• The model dynamically suggests tests during the diagnostic process.
• It supports decision-making for selecting the next diagnostic test.
• The proposed model demonstrated superior performance compared to traditional experience-based diagnostic models.

Conclusions:

• The developed Decision-Theoretic Model enhances sequential diagnosis by effectively managing uncertainty.
• It provides a structured approach for knowledge engineers to model diagnostic knowledge.
• The model offers a practical tool for prioritizing diagnostic tests, improving decision-making accuracy.