A logic programming approach to medical errors in imaging.

Area of Science:

• Medical Imaging
• Patient Safety
• Health Informatics

Background:

• Medical errors pose significant risks in healthcare delivery, necessitating robust adverse event reporting and learning systems.
• Current patient safety improvement efforts are slow due to a lack of systems that handle incomplete and uncertain real-world information.
• Effective root cause analysis requires models that can manage imprecise data and facilitate knowledge sharing.

Purpose of the Study:

• To develop and evaluate an adverse event reporting and learning system for medical imaging using Extended Logic Programming.
• To identify key causes of medical errors in imaging and define strategies for improvement.
• To generate reports on the impact, occurrence, and type of adverse events in healthcare institutions.

Main Methods:

• Extended Logic Programming was employed for knowledge representation and reasoning with incomplete and uncertain data.
• The Eindhoven Classification Model was adapted for classifying adverse event root causes in medical imaging.
• A multi-agent system approach was utilized for problem-solving and knowledge extraction.

Main Results:

• The developed system successfully classified adverse event root causes in the medical imaging domain.
• Deployment in two Portuguese healthcare institutions demonstrated that a majority of adverse events were concentrated in a few avoidable occurrences.
• The system facilitated automatic knowledge extraction for generating actionable reports.

Conclusions:

• The adapted Eindhoven Classification Model and Extended Logic Programming provide a robust framework for analyzing medical imaging adverse events.
• The developed adverse event reporting and learning system is effective in identifying and addressing preventable errors.
• The system contributes to improving the quality of care by enabling data-driven insights and targeted interventions.