Natural Language Processing for Literature Search in Vascular Surgery: A Pilot Study Testing an Artificial...

Area of Science:

• Vascular Surgery
• Medical Informatics
• Artificial Intelligence

Background:

• Literature searching in vascular surgery remains underexplored using Natural Language Processing (NLP).
• Current manual literature searches are time-consuming and may lack efficiency.

Purpose of the Study:

• To evaluate an AI-powered mobile application for literature searching in vascular surgery.
• To compare the AI application's performance against traditional expert-led literature searches.

Main Methods:

• A focused scientific question guided the literature search evaluation.
• An AI application performed automated literature searching, including quality assessment, relevancy filtering via NLP, summarization, and visualization.
• Results were benchmarked against a manual search conducted by human experts.

Main Results:

• The AI search completed in under an hour, compared to 10 hours for the manual search.
• The AI identified a higher proportion of relevant articles (48.6%) than the manual search (16.9%) (p < .010).
• The AI successfully excluded unrelated articles (p < .001), though it was not fully exhaustive.

Conclusions:

• AI-driven applications provide a targeted, efficient, and time-saving approach to literature searching in vascular surgery.
• These AI tools can serve as valuable complementary resources for researchers and clinicians.
• The AI application aids in continuous education and knowledge dissemination within the field.