ArticleBMC health services research2025
Harnessing ChatGPT for abstract screening in health-related scoping reviews: the role of structured eligibility criteria.
Article in BMC health services research, 2025. The graph could read no effect estimate from its abstract, so it casts no vote on the map. Not yet cited in PubMed.
What it found
Each row is one number read from the abstract, on the scale the paper reported it, with its interval. Left of the dashed line favours the treatment, right favours the comparator. Under each row is the sentence it came from. New to these charts? A ten-minute tutorial.
The abstract states no effect estimate the extractor could read, or names no intervention and outcome on the map, so this paper lights no cell and moves no belief. It is still indexed, cited and linked below.
The trial behind it
Trials whose registry record cites this paper, or whose number appears in the abstract. A trial that started after this paper was published is citing it as background, not reporting it.
Neither the registry nor the abstract names a trial number. If this is a trial report, that itself is worth knowing.
Who cites it
0 citing papers in PubMed.
No citing paper in PubMed yet.
Corrections and comments
PubMed lists nothing against this paper. Absence here is not a guarantee, only a check that was made.
Authors and funding
2 authors.
Funding
No grant is acknowledged in the PubMed record.
Abstract
backgroundScoping reviews, especially the screening phase, are time-consuming. Due to an exponential increase in the number of scientific articles related to health issues being published each year, the time burden is also expected to rise. In parallel, there is a gap in machine learning tools specifically tailored to support screening processes within scoping reviews. Large language models, such as ChatGPT, have shown promise in accelerating this process. Given the reliance of artificial intelligence on structured inputs, this study aims to refine eligibility criteria and analyse ChatGPT’s performance in abstract screening for a scoping review on digital tools supporting interprofessional interactions in healthcare.
methodsWe conducted a thematic analysis of ChatGPT 4.0’s explanations based on a previously conducted scoping review and developed three refined sets of eligibility criteria (narrow, wide, and balanced). ChatGPT reassessed these criteria using the human reviewers’ decisions as the gold standard. We calculated performance metrics such as sensitivity, specificity, and accuracy. Additionally, we combined decisions using majority voting and human conflict resolution to assess their combined performance.
resultsRephrasing the eligibility criteria using the Population-Concept-Context framework revealed challenges in each category: the complexity of healthcare provider interactions (population), ambiguities in defining digital tools (concept), and difficulties in identifying the healthcare setting (context). For example, ChatGPT was sometimes overinclusive, including individuals such as politicians, while at other times applying overly rigid boundaries, excluding healthcare groups like palliative care providers. The wide set of eligibility criteria achieved the highest sensitivity, while the narrow set had the highest specificity, with the original and balanced sets performing between these extremes. Combining decisions provided a good balance between sensitivity and specificity, effectively identifying ambiguous abstracts for manual review. The combination of balanced and narrow criteria resulted in a lower overall workload compared to other combinations.
conclusionRefining and structuring the eligibility criteria as well as combining different sets improved ChatGPT’s decision-making accuracy, highlighting the importance of well-defined and well-formulated criteria for such tasks. The formulation of eligibility criteria directly impacted screening outcomes: broad criteria maximized sensitivity at the expense of specificity, while narrow criteria produced the opposite effect. Combining different approaches may enable screening by one human reviewer accompanied by an AI-assisted screening for the major part and to include a second human reviewer in cases of discrepancies. However, challenges such as trust and transparency persist and need to be addressed for full integration into the review process.
Indexed as
Identifiers
What OpenQuestion holds
Registered trials
Read under generation 80e0d062 · epoch 390. Bibliography from PubMed, PubMed Central and OpenAlex; grants from NIH RePORTER; trial links from ClinicalTrials.gov; estimates, votes and beliefs from the OpenQuestion graph.