Human-Agentic AI Collaboration Workflow for Trend Analysis of ChatGPT Applications in Healthcare

Background: With the rapid advancement of artificial intelligence (AI), large language models (LLMs) such as ChatGPT have demonstrated strong potential to transform healthcare research and practice. However, LLMs are probabilistic and generate inherently variable outputs, which create risk and uncertainty in healthcare applications. Therefore, analyzing the current research landscape of ChatGPT applications is imperative to understand applied techniques and related concerns and to facilitate their practical integration into healthcare information systems. Objective: Leveraging LLM-based agentic AI with domain experts to analyze trends in ChatGPT healthcare applications and to examine the collaborative and interpretative capacities of LLM agents in broader contexts beyond narrowly defined topics. Methods: The review used a systematic search of the PubMed and Web of Science databases, following PRISMA-ScR guidelines, and focused on ChatGPT applications, associated concerns, applied techniques, and usage factors. To support the review process, an adaptive, iterative LLM-based agentic AI with a human-in-the-loop mechanism was employed for automated metadata extraction, identification, classification, and knowledge synthesis. The extracted information was subsequently used for temporal and thematic analyses, citation network construction, and main-path analysis. Results: Across 192 papers, the review identified 511 ChatGPT application types, 690 concern items, and 368 influence usage factors distinct from traditional IT, achieving a metadata error rate of 1.04%. The human-agentic AI workflow enabled efficient synthesis of contextual information across studies, addressing the growing challenge of timely evidence integration in medical knowledge. The findings provide insights into the evolving trajectories of ChatGPT applications in healthcare and demonstrate the value of an expert-validated human-agentic AI workflow in supporting literature analysis and streamlining review processes. Conclusion: This review provides a comprehensive overview of emerging ChatGPT applications, associated concerns, applied techniques, and usage factors in healthcare. The study further provides meaningful insights into the operational mechanisms of agentic AI, demonstrating the potential of human-AI collaboration for medical informatics research.

Authors

Institutions

Publication Details

Journal
medRxiv
Published
2026-11-17
DOI
https://doi.org/10.64898/2026.11.14.26368120
Primary Topic
Artificial Intelligence in Healthcare and Education
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Human-Agentic AI Collaboration Workflow for Trend Analysis of ChatGPT Applications in Healthcare

Eldon Y. Li, Yu Lin, Po Ju Chou
2 links
medRxiv
Artificial Intelligence in Healthcare and Education
preprint

Human-Agentic AI Collaboration Workflow for Trend Analysis of ChatGPT Applications in Healthcare

Eldon Y. Li, Yu Lin, Po Ju Chou
preprint en

Abstract

Background: With the rapid advancement of artificial intelligence (AI), large language models (LLMs) such as ChatGPT have demonstrated strong potential to transform healthcare research and practice. However, LLMs are probabilistic and generate inherently variable outputs, which create risk and uncertainty in healthcare applications. Therefore, analyzing the current research landscape of ChatGPT applications is imperative to understand applied techniques and related concerns and to facilitate their practical integration into healthcare information systems. Objective: Leveraging LLM-based agentic AI with domain experts to analyze trends in ChatGPT healthcare applications and to examine the collaborative and interpretative capacities of LLM agents in broader contexts beyond narrowly defined topics. Methods: The review used a systematic search of the PubMed and Web of Science databases, following PRISMA-ScR guidelines, and focused on ChatGPT applications, associated concerns, applied techniques, and usage factors. To support the review process, an adaptive, iterative LLM-based agentic AI with a human-in-the-loop mechanism was employed for automated metadata extraction, identification, classification, and knowledge synthesis. The extracted information was subsequently used for temporal and thematic analyses, citation network construction, and main-path analysis. Results: Across 192 papers, the review identified 511 ChatGPT application types, 690 concern items, and 368 influence usage factors distinct from traditional IT, achieving a metadata error rate of 1.04%. The human-agentic AI workflow enabled efficient synthesis of contextual information across studies, addressing the growing challenge of timely evidence integration in medical knowledge. The findings provide insights into the evolving trajectories of ChatGPT applications in healthcare and demonstrate the value of an expert-validated human-agentic AI workflow in supporting literature analysis and streamlining review processes. Conclusion: This review provides a comprehensive overview of emerging ChatGPT applications, associated concerns, applied techniques, and usage factors in healthcare. The study further provides meaningful insights into the operational mechanisms of agentic AI, demonstrating the potential of human-AI collaboration for medical informatics research.

medRxiv
National Chung Cheng University (TW)
Good health and well-being
Artificial Intelligence in Healthcare and Education
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon