Advancing Artificial Intelligence for Clinical Knowledge Retrieval: A Case Study Using ChatGPT-4 and Link Retrieval...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare

Background:

• Accurate retrieval of clinical information from medical guidelines is crucial for healthcare professionals.
• Large language models like ChatGPT-4 offer potential but require validation for medical applications.

Purpose of the Study:

• To enhance the traceability and retrieval accuracy of ChatGPT-4 for medical text.
• To systematically retrieve clinical answers from international diabetic ketoacidosis (DKA) guidelines.

Main Methods:

• Developed a systematic methodology for AI-guided information retrieval.
• Utilized ChatGPT-4 with the 'Link Reader' plug-in to access and extract information from DKA guidelines.
• Iteratively asked one question per guideline, compiled answers with citations, and repeated 30 times for consistency.

Main Results:

• Integration of ChatGPT-4 and 'Link Reader' significantly improved traceability and retrieval accuracy.
• AI model successfully provided accurate clinical answers from guidelines, despite minor challenges.
• The systematic approach demonstrated promising results for automated medical text synthesis.

Conclusions:

• AI text-generation models, when systematically applied, are valuable tools for medical professionals.
• The developed framework enhances AI model traceability and accuracy for medical information synthesis.
• Further advancements can improve AI-generated recommendations in medicine and scientific research.