Utilizing Large language models to select literature for meta-analysis shows workload reduction while maintaining a similar recall level as manual curation.
Xiangming Cai, Yuanming Geng, Yiming Du, Bart Westerman, Duolao Wang, Chiyuan Ma, Juan J Garcia Vallejo
{"title":"Utilizing Large language models to select literature for meta-analysis shows workload reduction while maintaining a similar recall level as manual curation.","authors":"Xiangming Cai, Yuanming Geng, Yiming Du, Bart Westerman, Duolao Wang, Chiyuan Ma, Juan J Garcia Vallejo","doi":"10.1186/s12874-025-02569-3","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>Large language models (LLMs) like ChatGPT showed great potential in aiding medical research. A heavy workload in filtering records is needed during the research process of evidence-based medicine, especially meta-analysis. However, few studies tried to use LLMs to help screen records in meta-analysis.</p><p><strong>Objective: </strong>In this research, we aimed to explore the possibility of incorporating multiple LLMs to facilitate the screening step based on the title and abstract of records during meta-analysis.</p><p><strong>Methods: </strong>Various LLMs were evaluated, which includes GPT-3.5, GPT-4, Deepseek-R1-Distill, Qwen-2.5, Phi-4, Llama-3.1, Gemma-2 and Claude-2. To assess our strategy, we selected three meta-analyses from the literature, together with a glioma meta-analysis embedded in the study, as additional validation. For the automatic selection of records from curated meta-analyses, a four-step strategy called LARS-GPT was developed, consisting of (1) criteria selection and single-prompt (prompt with one criterion) creation, (2) best combination identification, (3) combined-prompt (prompt with one or more criteria) creation, and (4) request sending and answer summary. Recall, workload reduction, precision, and F1 score were calculated to assess the performance of LARS-GPT.</p><p><strong>Results: </strong>A variable performance was found between different single-prompts, with a mean recall of 0.800. Based on these single-prompts, we were able to find combinations with better performance than the pre-set threshold. Finally, with a best combination of criteria identified, LARS-GPT showed a 40.1% workload reduction on average with a recall greater than 0.9.</p><p><strong>Conclusions: </strong>We show here the groundbreaking finding that automatic selection of literature for meta-analysis is possible with LLMs. We provide it here as a pipeline, LARS-GPT, which showed a great workload reduction while maintaining a pre-set recall.</p>","PeriodicalId":9114,"journal":{"name":"BMC Medical Research Methodology","volume":"25 1","pages":"116"},"PeriodicalIF":3.9000,"publicationDate":"2025-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12036192/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"BMC Medical Research Methodology","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1186/s12874-025-02569-3","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"HEALTH CARE SCIENCES & SERVICES","Score":null,"Total":0}
引用次数: 0
Abstract
Background: Large language models (LLMs) like ChatGPT showed great potential in aiding medical research. A heavy workload in filtering records is needed during the research process of evidence-based medicine, especially meta-analysis. However, few studies tried to use LLMs to help screen records in meta-analysis.
Objective: In this research, we aimed to explore the possibility of incorporating multiple LLMs to facilitate the screening step based on the title and abstract of records during meta-analysis.
Methods: Various LLMs were evaluated, which includes GPT-3.5, GPT-4, Deepseek-R1-Distill, Qwen-2.5, Phi-4, Llama-3.1, Gemma-2 and Claude-2. To assess our strategy, we selected three meta-analyses from the literature, together with a glioma meta-analysis embedded in the study, as additional validation. For the automatic selection of records from curated meta-analyses, a four-step strategy called LARS-GPT was developed, consisting of (1) criteria selection and single-prompt (prompt with one criterion) creation, (2) best combination identification, (3) combined-prompt (prompt with one or more criteria) creation, and (4) request sending and answer summary. Recall, workload reduction, precision, and F1 score were calculated to assess the performance of LARS-GPT.
Results: A variable performance was found between different single-prompts, with a mean recall of 0.800. Based on these single-prompts, we were able to find combinations with better performance than the pre-set threshold. Finally, with a best combination of criteria identified, LARS-GPT showed a 40.1% workload reduction on average with a recall greater than 0.9.
Conclusions: We show here the groundbreaking finding that automatic selection of literature for meta-analysis is possible with LLMs. We provide it here as a pipeline, LARS-GPT, which showed a great workload reduction while maintaining a pre-set recall.
期刊介绍：
BMC Medical Research Methodology is an open access journal publishing original peer-reviewed research articles in methodological approaches to healthcare research. Articles on the methodology of epidemiological research, clinical trials and meta-analysis/systematic review are particularly encouraged, as are empirical studies of the associations between choice of methodology and study outcomes. BMC Medical Research Methodology does not aim to publish articles describing scientific methods or techniques: these should be directed to the BMC journal covering the relevant biomedical subject area.
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号
