AI-driven literature tools for enhanced medical education and research

Abstract

Objective: To enhance medical education and research by developing and evaluating AI-powered tools that improve literature search efficiency, facilitate comprehensive reviews, and increase engagement with current literature among clinicians and trainees. Hypothesis: Integrating generative AI into literature search and review processes can significantly enhance the accessibility, efficiency, and engagement with medical literature for educators, clinicians, and trainees. Methods: We developed an integrated suite of AI-driven tools: 1. Initial Literature Search Tool: Utilizes generative AI to construct and iteratively refine PubMed queries, ensuring retrieval of comprehensive and relevant articles. The tool automates downloading abstracts and permissible full-text PDFs, providing assured-real citations directly from PubMed to prevent AI-generated errors. 2. Scoping Review Assistant: Guides users through scoping reviews with AI-powered suggestions and automation at each step, including search iteration, article categorization, summarization, drafting assistance, and bibliography generation. 3. Automated Monthly Literature Digest: Generates monthly summaries of new publications in anesthesiology subspecialties using AI to compile, categorize, and summarize recent articles, delivered via an interactive web interface. 3. AI-Generated Podcast Tool: Converts research articles into customized audio summaries tailored to different audience levels, allowing users to specify target topics, focus areas, and length preferences. 4. All tools are built on a common backend with web interfaces, emphasizing user interaction and compliance with copyright regulations. Data: Engagement metrics for the literature digest were tracked over four months, capturing anonymous click rates across anesthesiology subspecialties. Preliminary user feedback was collected for the podcast tool, focusing on usability and educational value. No human subject research was conducted; feedback reflects authors' experiences. Results: The literature digest tool demonstrated high engagement, indicating strong interest among clinicians in subspecialties like pain management, perioperative medicine, cardiac, obstetric, regional, general, and critical care anesthesia. Users appreciated the efficient delivery of relevant information. The podcast tool received positive feedback for enhancing accessibility; users found it beneficial for supplementing learning during commutes and appreciated customization options. The scoping review assistant and initial search tool effectively streamlined literature search and review processes, reducing manual workload and improving comprehensiveness. Conclusions: AI-driven literature tools significantly enhance medical education and research by improving access to and engagement with current literature. These tools support clinicians and trainees in staying updated, integrating evidence-based practices, and facilitating knowledge synthesis. Future developments will focus on refining AI models for greater specificity, assessing long-term educational outcomes, and exploring integration with clinical information systems to further support medical education and practice. This abstract was presented at the American Physiology Summit 2025 and is only available in HTML format. There is no downloadable file or PDF version. The Physiology editorial board was not involved in the peer review process.

Ähnliche Arbeiten