Synthesis of elementary single-disease recommendations to support guideline-based therapeutic decision for complex...

Area of Science:

• Medical Informatics
• Clinical Decision Support Systems
• Artificial Intelligence in Healthcare

Background:

• Clinical practice guidelines (CPGs) typically address single diseases, lacking support for patients with multiple comorbidities.
• Existing computer-based systems often handle complex patient cases, but integration with CPGs can be challenging.

Purpose of the Study:

• To develop a module for automatically generating IF-THEN-WITH decision rules from GEM-encoded CPGs.
• To enhance decision support for physicians managing patients with multiple diseases.

Main Methods:

• Utilized GEM-encoded instances of CPGs to develop a rule generation module.
• Implemented a two-stage unification process for rule synthesis, eliminating redundancies and incoherences.
• Triggered rules based on partial matching with patient clinical profiles.

Main Results:

• Successfully generated IF-THEN-WITH decision rules from CPGs.
• A synthesis process effectively managed rule redundancies and incoherences.
• Physicians were presented with remaining, potentially contradictory, recommendations for final decision-making.

Conclusions:

• The developed system provides valuable decision support for complex patients with multiple diseases.
• Physician oversight is maintained, allowing them to manage therapeutic controversies.
• This approach bridges the gap between CPGs and the management of multimorbidity.