{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,5]],"date-time":"2026-10-05T16:09:48Z","timestamp":1791216588482,"version":"4.1.0"},"reference-count":31,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,5,23]],"date-time":"2025-05-23T00:00:00Z","timestamp":1747958400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["82201148"],"award-info":[{"award-number":["82201148"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004731","name":"Natural Science Foundation of Zhejiang Province","doi-asserted-by":"publisher","award":["LQ22H120002"],"award-info":[{"award-number":["LQ22H120002"]}],"id":[{"id":"10.13039\/501100004731","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100005315","name":"Natural Science Foundation of Ningbo Municipality","doi-asserted-by":"publisher","award":["2023J390"],"award-info":[{"award-number":["2023J390"]}],"id":[{"id":"10.13039\/501100005315","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Cell Dev. Biol."],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>The development of medical artificial intelligence (AI) models is primarily driven by the need to address healthcare resource scarcity, particularly in underserved regions. Proposing an affordable, accessible, interpretable, and automated AI system for non-clinical settings is crucial to expanding access to quality healthcare.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>This cross-sectional study developed the Multimodal Ocular Surface Assessment and Interpretation Copilot (MOSAIC) using three multimodal large language models: gpt-4-turbo, claude-3-opus, and gemini-1.5-pro-latest, for detecting three ocular surface diseases (OSDs) and grading keratitis and pterygium. A total of 375 smartphone-captured ocular surface images collected from 290 eyes were utilized to validate MOSAIC. The performance of MOSAIC was evaluated in both zero-shot and few-shot settings, with tasks including image quality control, OSD detection, analysis of the severity of keratitis, and pterygium grading. The interpretability of the system was also evaluated.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>MOSAIC achieved 95.00% accuracy in image quality control, 86.96% in OSD detection, 88.33% in distinguishing mild from severe keratitis, and 66.67% in determining pterygium grades with five-shot settings. The performance significantly improved with the increasing learning shots (p &amp;lt; 0.01). The system attained high ROUGE-L F1 scores of 0.70\u20130.78, depicting its interpretable image comprehension capability.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>MOSAIC exhibited exceptional few-shot learning capabilities, achieving high accuracy in OSD management with minimal training examples. This system has significant potential for smartphone integration to enhance the accessibility and effectiveness of OSD detection and grading in resource-limited settings.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fcell.2025.1600202","type":"journal-article","created":{"date-parts":[[2025,5,23]],"date-time":"2025-05-23T01:29:53Z","timestamp":1747963793000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":19,"title":["Large language model-based multimodal system for detecting and grading ocular surface diseases from smartphone images"],"prefix":"10.3389","volume":"13","author":[{"given":"Zhongwen","family":"Li","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhouqian","family":"Wang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Liheng","family":"Xiu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Pengyao","family":"Zhang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wenfang","family":"Wang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yangyang","family":"Wang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Gang","family":"Chen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Weihua","family":"Yang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wei","family":"Chen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2025,5,23]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"1721","DOI":"10.1001\/jama.2013.280318","article-title":"Conjunctivitis: a systematic review of diagnosis and treatment","volume":"310","author":"Azari","year":"2013","journal-title":"JAMA"},{"key":"B2","unstructured":"Language models are few-shot learners\n          \n          \n            \n              Brown\n              T. B.\n            \n            \n              Mann\n              B.\n            \n            \n              Ryder\n              N.\n            \n            \n              Subbiah\n              M.\n            \n            \n              Kaplan\n              J.\n            \n            \n              Dhariwal\n              P.\n            \n          \n          10.48550\/arXiv.2005.14165\n          \n          2020"},{"key":"B3","first-page":"33","article-title":"Prevention, treatment and rehabilitation","volume":"22","author":"Burton","year":"2009","journal-title":"Community Eye Health"},{"key":"B4","doi-asserted-by":"publisher","first-page":"e489","DOI":"10.1016\/S2214-109X(20)30488-5","article-title":"The lancet global health commission on global eye health: vision beyond 2020","volume":"9","author":"Burton","year":"2021","journal-title":"Lancet Glob. Health"},{"key":"B5","doi-asserted-by":"publisher","first-page":"106","DOI":"10.1177\/1357633X17742182","article-title":"Models of care in tele-ophthalmology: a scoping review","volume":"25","author":"Caffery","year":"2019","journal-title":"J. Telemed. Telecare"},{"key":"B6","unstructured":"Closing the translation gap: AI applications in digital pathology - PubMed\n          \n          \n          2025"},{"key":"B8","doi-asserted-by":"publisher","first-page":"198","DOI":"10.4103\/0970-0218.120153","article-title":"Burden of corneal blindness in India","volume":"38","author":"Gupta","year":"2013","journal-title":"Indian J. Community Med."},{"key":"B9","doi-asserted-by":"publisher","first-page":"205","DOI":"10.1038\/s41746-024-01208-3","article-title":"Evaluating multimodal AI in medical diagnostics","volume":"7","author":"Kaczmarczyk","year":"2024","journal-title":"npj Digit. Med."},{"key":"B10","doi-asserted-by":"publisher","first-page":"4705","DOI":"10.1038\/s41467-024-48998-4","article-title":"ChatMOF: an artificial intelligence system for predicting and generating metal-organic frameworks using large language models","volume":"15","author":"Kang","year":"2024","journal-title":"Nat. Commun."},{"key":"B11","doi-asserted-by":"publisher","first-page":"531","DOI":"10.1097\/OPX.0b013e31817dba2e","article-title":"Grading contact lens-related microbial keratitis: relevance to disease burden","volume":"85","author":"Keay","year":"2008","journal-title":"Optom. Vis. Sci."},{"key":"B12","doi-asserted-by":"publisher","first-page":"3738","DOI":"10.1038\/s41467-021-24116-6","article-title":"Preventing corneal blindness caused by keratitis using artificial intelligence","volume":"12","author":"Li","year":"","journal-title":"Nat. Commun."},{"key":"B13","doi-asserted-by":"publisher","first-page":"106048","DOI":"10.1016\/j.cmpb.2021.106048","article-title":"Development of a deep learning-based image quality control system to detect and filter out ineligible slit-lamp images: a multicenter study","volume":"203","author":"Li","year":"","journal-title":"Comput. Methods Programs Biomed."},{"key":"B14","doi-asserted-by":"publisher","first-page":"104722","DOI":"10.1016\/j.jbi.2024.104722","article-title":"Promoting smartphone-based keratitis screening using meta-learning: a multicenter study","volume":"157","author":"Li","year":"2024","journal-title":"J. Biomed. Inf."},{"key":"B15","doi-asserted-by":"publisher","first-page":"195","DOI":"10.1145\/3560815","article-title":"Pre-train, prompt, and predict: a systematic survey of prompting methods in natural language processing","volume":"55","author":"Liu","year":"2023","journal-title":"ACM Comput. Surv."},{"key":"B16","doi-asserted-by":"publisher","first-page":"336","DOI":"10.1136\/bjo-2022-322552","article-title":"Accurate detection and grading of pterygium through smartphone by a fusion training model","volume":"108","author":"Liu","year":"2024","journal-title":"Br. J. Ophthalmol."},{"key":"B17","doi-asserted-by":"publisher","first-page":"4096","DOI":"10.1097\/JS9.0000000000001359","article-title":"Step into the era of large multimodal models: a pilot study on ChatGPT-4V(ision)\u2019s ability to interpret radiological images","volume":"110","author":"Lingxuan","year":"2024","journal-title":"Int. J. Surg. Lond. Engl."},{"key":"B18","doi-asserted-by":"publisher","first-page":"383","DOI":"10.4103\/0301-4738.33829","article-title":"Pterygium-induced corneal refractive changes","volume":"55","author":"Maheshwari","year":"2007","journal-title":"Indian J. Ophthalmol."},{"key":"B19","doi-asserted-by":"publisher","first-page":"783","DOI":"10.1136\/bjophthalmol-2011-301378","article-title":"The number of ophthalmologists in practice and training worldwide: a growing gap despite more than 200,000 practitioners","volume":"96","author":"Resnikoff","year":"2012","journal-title":"Br. J. Ophthalmol."},{"key":"B20","doi-asserted-by":"publisher","first-page":"719","DOI":"10.1016\/j.survophthal.2018.03.001","article-title":"Prevalence and risk factors of pterygium: a systematic review and meta-analysis","volume":"63","author":"Rezvan","year":"2018","journal-title":"Surv. Ophthalmol."},{"key":"B21","doi-asserted-by":"publisher","first-page":"253","DOI":"10.1016\/s0161-6420(02)01839-0","article-title":"Vision loss: a public health problem?","volume":"110","author":"Saaddine","year":"2003","journal-title":"Ophthalmology"},{"key":"B22","doi-asserted-by":"publisher","first-page":"10785","DOI":"10.1038\/s41598-024-60405-y","article-title":"Evaluating the strengths and weaknesses of large language models in answering neurophysiology questions","volume":"14","author":"Shojaee-Mend","year":"2024","journal-title":"Sci. Rep."},{"key":"B23","doi-asserted-by":"publisher","first-page":"351","DOI":"10.1016\/j.jtos.2021.08.007","article-title":"The epidemiology of infectious keratitis","volume":"28","author":"Stapleton","year":"2023","journal-title":"Ocul. Surf."},{"key":"B24","doi-asserted-by":"publisher","first-page":"1516","DOI":"10.1016\/j.ophtha.2012.01.052","article-title":"Risk factors for moderate and severe microbial keratitis in daily wear contact lens users","volume":"119","author":"Stapleton","year":"2012","journal-title":"Ophthalmology"},{"key":"B25","doi-asserted-by":"publisher","first-page":"988","DOI":"10.1002\/jrsm.1749","article-title":"Zero- and few-shot prompting of generative large language models provides weak assessment of risk of bias in clinical trials","volume":"15","author":"\u0160uster","year":"2024","journal-title":"Res. Synth. Methods"},{"key":"B26","doi-asserted-by":"publisher","first-page":"1299","DOI":"10.1109\/TMI.2016.2535302","article-title":"Convolutional neural networks for medical image analysis: full training or fine tuning?","volume":"35","author":"Tajbakhsh","year":"2016","journal-title":"IEEE Trans. Med. Imaging"},{"key":"B27","doi-asserted-by":"publisher","first-page":"e1432","DOI":"10.1016\/S2214-109X(23)00323-6","article-title":"Artificial intelligence and digital health in global eye health: opportunities and challenges","volume":"11","author":"Tan","year":"2023","journal-title":"Lancet Glob. Health"},{"key":"B28","doi-asserted-by":"publisher","first-page":"728","DOI":"10.1186\/s12967-023-04576-8","article-title":"Harnessing large language models (LLMs) for candidate gene prioritization and selection","volume":"21","author":"Toufiq","year":"2023","journal-title":"J. Transl. Med."},{"key":"B29","doi-asserted-by":"publisher","first-page":"41","DOI":"10.1038\/s41746-024-01029-4","article-title":"Prompt engineering in consistency and reliability with the evidence-based guideline for LLMs","volume":"7","author":"Wang","year":"2024","journal-title":"NPJ Digit. Med."},{"key":"B30","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2302.11382","article-title":"A prompt pattern catalog to enhance prompt engineering with ChatGPT","author":"White","year":"2023"},{"key":"B31","doi-asserted-by":"publisher","first-page":"1133680","DOI":"10.3389\/fcell.2023.1133680","article-title":"Artificial intelligence-assisted diagnosis of ocular surface diseases","volume":"11","author":"Zhang","year":"2023","journal-title":"Front. Cell. Dev. Biol."},{"key":"B7","doi-asserted-by":"publisher","first-page":"181","DOI":"10.1038\/s41746-024-01174-w","article-title":"Deep learning for multi-type infectious keratitis diagnosis: A nationwide, cross-sectional, multicenter study","volume":"7","author":"Zhongwen","year":"2025","journal-title":"npj Digit. Med."}],"container-title":["Frontiers in Cell and Developmental Biology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcell.2025.1600202\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,5,23]],"date-time":"2025-05-23T01:29:56Z","timestamp":1747963796000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcell.2025.1600202\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,5,23]]},"references-count":31,"alternative-id":["10.3389\/fcell.2025.1600202"],"URL":"https:\/\/doi.org\/10.3389\/fcell.2025.1600202","relation":{},"ISSN":["2296-634X"],"issn-type":[{"value":"2296-634X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,5,23]]},"article-number":"1600202"}}