{"database":"biostudies-literature","file_versions":[],"scores":null,"additional":{"omics_type":["Unknown"],"volume":["13"],"submitter":["Huang M"],"pubmed_abstract":["<h4>Background</h4>Large language models (LLMs) are increasingly accessed by lay users for medical advice. This study aims to conduct a comprehensive evaluation of the responses generated by five large language models.<h4>Methods</h4>We identified 31 ophthalmology-related questions most frequently raised by patients during routine consultations and subsequently elicited responses from five large language models: ChatGPT-4o, DeepSeek-V3, Doubao, Wenxin Yiyan 4.0 Turbo, and Qwen. A five-point likert scale was employed to assess each model across five domains: accuracy, logical consistency, coherence, safety, and content accessibility. Additionally, textual characteristics, including character, word, and sentence counts, were quantitatively analyzed.<h4>Results</h4>ChatGPT-4o and DeepSeek-V3 "],"journal":["Frontiers in public health"],"pagination":["1673045"],"full_dataset_link":["https://www.ebi.ac.uk/biostudies/studies/S-EPMC12497754"],"repository":["biostudies-literature"],"pubmed_title":["Comparative performance of large language models for patient-initiated ophthalmology consultations."],"pmcid":["PMC12497754"],"pubmed_authors":["Yang W","Zhang Z","Zhou S","Huang M","Cui X","Wang X","Chi W","Xu Y"],"additional_accession":[]},"is_claimable":false,"name":"Comparative performance of large language models for patient-initiated ophthalmology consultations.","description":"<h4>Background</h4>Large language models (LLMs) are increasingly accessed by lay users for medical advice. This study aims to conduct a comprehensive evaluation of the responses generated by five large language models.<h4>Methods</h4>We identified 31 ophthalmology-related questions most frequently raised by patients during routine consultations and subsequently elicited responses from five large language models: ChatGPT-4o, DeepSeek-V3, Doubao, Wenxin Yiyan 4.0 Turbo, and Qwen. A five-point likert scale was employed to assess each model across five domains: accuracy, logical consistency, coherence, safety, and content accessibility. Additionally, textual characteristics, including character, word, and sentence counts, were quantitatively analyzed.<h4>Results</h4>ChatGPT-4o and DeepSeek-V3 ","dates":{"release":"2025-01-01T00:00:00Z","publication":"2025","modification":"2026-05-06T03:19:28.794Z","creation":"2026-05-06T03:12:35.03Z"},"accession":"S-EPMC12497754","cross_references":{"pubmed":["41059182"],"doi":["10.3389/fpubh.2025.1673045"]}}