{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,1]],"date-time":"2026-08-01T00:43:56Z","timestamp":1785545036162,"version":"3.56.0"},"reference-count":35,"publisher":"JMIR Publications Inc.","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["JMIR Med Inform"],"abstract":"Abstract<\/jats:title>\n \n Background<\/jats:title>\n Patients undergoing invasive procedures frequently experience anxiety and often have unanswered questions regarding the procedure. Although large language models show considerable promise for supporting patient communication in many cases, their deployment in health care is limited by the risk of hallucinations, data-privacy constraints, and high energy costs\u2014factors that impede equitable access in resource-limited settings.<\/jats:p>\n <\/jats:sec>\n \n Objective<\/jats:title>\n This study aims to develop and evaluate LENOHA (Low Energy, No Hallucination, Leave No One Behind Architecture), a locally executable dialog system for safe, equitable, and sustainable preprocedural communication.<\/jats:p>\n <\/jats:sec>\n \n Methods<\/jats:title>\n \n We built expert-curated FAQ (frequently asked question) databases and independent test sets for 2 domains (tooth extraction and gastroscopy; 200 utterances per domain: 100 clinical questions and 100 casual). A sentence-transformer classifier routed inputs: clinical questions were answered verbatim from the vetted FAQs (nongenerative path), while casual conversation was handled by a locally hosted 8-billion-parameter small language model (Swallow-8B). We evaluated 4 sentence-transformer models (including E5-large-instruct) against cloud large language models (ChatGPT [GPT-4o] and Gemini Advanced) using accuracy,\n F<\/jats:italic>\n 1<\/jats:sub>\n -score, and area under the receiver operating characteristic curve, and measured the on-device inference energy on a consumer graphics processing unit (RTX 3080).\n <\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n \n Across both domains (N=400), E5-large-instruct achieved an accuracy of 98.3% (393\/400; 95% CI 96.4%\u201099.1%) and an area under the curve of 0.996, with only 7 out of 400 (1.8%) misclassifications. This performance was not statistically different from that of ChatGPT (GPT-4o), which had 6 out of 400 (1.5%) errors (McNemar test with Holm adjustment;\n P<\/jats:italic>\n >.99). Sustainability measurements showed approximately 2.23 mWh per request (latency\u22480.10 s; video RAM\u22482.2 GiB average, \u22482.5 GiB peak) for the nongenerative clinical path vs approximately 168.27 mWh (latency\u22488.51 s; video RAM\u224813.3 GiB average, \u224814.0 GiB peak) for small language model small talk\u2014approximately a 75-fold higher energy footprint per reply for the generative path.\n <\/jats:p>\n <\/jats:sec>\n \n Conclusions<\/jats:title>\n High-precision, nongenerative clinical support is feasible using local, low-cost hardware without cloud dependence. By decoupling clinical information retrieval from generative chitchat, LENOHA enhances safety, preserves privacy, and markedly reduces energy use, offering a practical blueprint for sustainable and equitable medical AI deployment across diverse care settings.<\/jats:p>\n <\/jats:sec>","DOI":"10.2196\/89173","type":"journal-article","created":{"date-parts":[[2026,7,21]],"date-time":"2026-07-21T13:47:06Z","timestamp":1784641626000},"page":"e89173-e89173","source":"Crossref","is-referenced-by-count":0,"title":["A Locally Executable AI System for Improving Preoperative Patient Communication: Multidomain Clinical Evaluation"],"prefix":"10.2196","volume":"14","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-0420-6183","authenticated-orcid":false,"given":"Motoki","family":"Sato","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0005-3277-1816","authenticated-orcid":false,"given":"Sou","family":"Nagata","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-5195-4663","authenticated-orcid":false,"given":"Mizuho","family":"Ohnuma","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1023-9053","authenticated-orcid":false,"given":"Hidekazu","family":"Takahashi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0003-7368-2612","authenticated-orcid":false,"given":"Tomoaki","family":"Kakazu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9053-8653","authenticated-orcid":false,"given":"Masayuki","family":"Yamamura","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7020-5085","authenticated-orcid":false,"given":"Atsushi","family":"Yoshikawa","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9577-3249","authenticated-orcid":false,"given":"Yuki","family":"Matsushita","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1010","published-online":{"date-parts":[[2026,7,21]]},"reference":[{"issue":"4","key":"R1","doi-asserted-by":"publisher","first-page":"368","DOI":"10.1111\/j.1834-7819.2010.01256.x","article-title":"The extent and nature of dental fear and phobia in Australia","volume":"55","author":"Armfield","journal-title":"Aust Dent J"},{"issue":"5","key":"R2","doi-asserted-by":"publisher","first-page":"293","DOI":"10.1016\/0266-4356(94)90049-3","article-title":"Patients\u2019 anxieties with third molar surgery","volume":"32","author":"Earl","journal-title":"Br J Oral Maxillofac Surg"},{"issue":"1","key":"R3","doi-asserted-by":"publisher","first-page":"151","DOI":"10.1177\/0272989X10364247","article-title":"Interventions to improve patient comprehension in informed consent for medical and surgical procedures: a systematic review","volume":"31","author":"Schenker","journal-title":"Med Decis Making"},{"issue":"1","key":"R4","doi-asserted-by":"publisher","DOI":"10.1093\/bjsopen\/zrac159","article-title":"Digital technology in informed consent for surgery: systematic review","volume":"7","author":"Kiernan","journal-title":"BJS Open"},{"issue":"3","key":"R5","doi-asserted-by":"publisher","first-page":"166","DOI":"10.2185\/jrm.2023-047","article-title":"Labor shortage of physicians in rural areas and surgical specialties caused by Work Style Reform Policies of the Japanese government: a quantitative simulation analysis","volume":"19","author":"Numata","journal-title":"J Rural Med"},{"issue":"4","key":"R6","doi-asserted-by":"publisher","first-page":"715","DOI":"10.1038\/s41391-023-00737-4","article-title":"Digital informed consent for urological surgery: randomized controlled study comparing multimedia-supported vs. traditional paper-based informed consent concerning satisfaction, anxiety, information gain and time efficiency","volume":"27","author":"Haack","journal-title":"Prostate Cancer Prostatic Dis"},{"issue":"1","key":"R7","doi-asserted-by":"publisher","first-page":"27","DOI":"10.1007\/s41669-024-00536-0","article-title":"Digital versus paper-based consent from the UK NHS perspective: a micro-costing analysis","volume":"9","author":"Houten","journal-title":"Pharmacoecon Open"},{"issue":"6","key":"R8","doi-asserted-by":"publisher","DOI":"10.3390\/healthcare11060887","article-title":"ChatGPT utility in healthcare education, research, and practice: systematic review on the promising perspectives and valid concerns","volume":"11","author":"Sallam","journal-title":"Healthcare (Basel)"},{"key":"R9","doi-asserted-by":"publisher","DOI":"10.2196\/59851","article-title":"Quality of conventional versus artificial intelligence oral surgery consent forms: comparative analysis","volume":"28","author":"Gaessler","journal-title":"J Med Internet Res"},{"issue":"1","key":"R10","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01576-4","article-title":"Mitigating the risk of health inequity exacerbated by large language models","volume":"8","author":"Ji","journal-title":"NPJ Digit Med"},{"issue":"3","key":"R11","doi-asserted-by":"publisher","first-page":"227","DOI":"10.1016\/j.adaj.2023.11.004","article-title":"Is ChatGPT a reliable source of scientific information regarding third-molar surgery?","volume":"155","author":"Aguiar de Sousa","journal-title":"J Am Dent Assoc"},{"key":"R12","unstructured":"Xu Z Jain S Kankanhalli M . Hallucination is inevitable: an innate limitation of large language models. arXiv. Preprint posted online on Jan 22, 2024. [doi: 10.48550\/arXiv.2401.11817]"},{"issue":"1","key":"R13","doi-asserted-by":"publisher","DOI":"10.1056\/AIra2400380","article-title":"RAG in health care: a novel framework for improving communication and decision-making by addressing LLM limitations","volume":"2","author":"Ng","journal-title":"NEJM AI"},{"key":"R14","doi-asserted-by":"publisher","DOI":"10.1162\/tacl_a_00638","article-title":"Lost in the middle: how language models use long contexts","volume":"12","author":"Liu","journal-title":"Trans Assoc Comput Linguist"},{"key":"R15","unstructured":"Introducing the model context protocol. Anthropic. URL: https:\/\/www.anthropic.com\/news\/model-context-protocol [Accessed 24-02-2026]"},{"key":"R16","doi-asserted-by":"publisher","DOI":"10.1145\/3796519","article-title":"Model context protocol (MCP): landscape, security threats, and future research directions","author":"Hou","journal-title":"ACM Trans Softw Eng Methodol"},{"issue":"17","key":"R17","doi-asserted-by":"publisher","first-page":"3190","DOI":"10.3390\/buildings15173190","article-title":"Human\u2013AI teaming in structural analysis: a model context protocol approach for explainable and accurate generative AI","volume":"15","author":"Avila","journal-title":"Buildings"},{"key":"R18","doi-asserted-by":"publisher","DOI":"10.1177\/1759720X251331529","article-title":"RAGing ahead in rheumatology: new language model architectures to tame artificial intelligence","volume":"17","author":"Benavent","journal-title":"Ther Adv Musculoskelet Dis"},{"issue":"3","key":"R19","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1007\/s11023-025-09730-3","article-title":"Global health in the age of AI: charting a course for ethical implementation and societal benefit","volume":"35","author":"Morley","journal-title":"Minds Mach"},{"issue":"11","key":"R20","doi-asserted-by":"publisher","first-page":"453","DOI":"10.1016\/j.diii.2024.06.002","article-title":"Climate change and artificial intelligence in healthcare: review and recommendations towards a sustainable future","volume":"105","author":"Ueda","journal-title":"Diagn Interv Imaging"},{"key":"R21","unstructured":"Adedeji A Sanni M Ayodele E Joshi S Olatunji T . The multicultural medical assistant: can LLMs improve medical ASR errors across borders? arXiv. Preprint posted online on Jan 25, 2025. [doi: 10.48550\/arXiv.2501.15310]"},{"key":"R22","unstructured":"Dichristofano A Shuster H Chandra S Patwari N . Global performance disparities between English-language accents in automatic speech recognition. arXiv. Preprint posted online on Aug 1, 2022. [doi: 10.48550\/arXiv.2208.01157]"},{"issue":"1","key":"R23","doi-asserted-by":"publisher","first-page":"e12","DOI":"10.1016\/S2589-7500(23)00225-X","article-title":"Assessing the potential of GPT-4 to perpetuate racial and gender biases in health care: a model evaluation study","volume":"6","author":"Zack","journal-title":"Lancet Digit Health"},{"key":"R24","doi-asserted-by":"publisher","DOI":"10.1016\/j.resconrec.2025.108670","article-title":"Assessing the carbon footprint of language models: towards sustainability in AI","volume":"226","author":"Jeanquartier","journal-title":"Resour Conserv Recycl"},{"issue":"11","key":"R25","doi-asserted-by":"publisher","first-page":"e848","DOI":"10.1016\/S2589-7500(24)00143-2","article-title":"Generative artificial intelligence and ethical considerations in health care: a scoping review and ethics checklist","volume":"6","author":"Ning","journal-title":"Lancet Digit Health"},{"key":"R26","doi-asserted-by":"publisher","DOI":"10.2196\/80028","article-title":"Scalable agile framework for execution in AI for medical AI ethics policy design in small- and medium-sized enterprises","volume":"28","author":"Nemteanu","journal-title":"J Med Internet Res"},{"issue":"1","key":"R27","doi-asserted-by":"publisher","DOI":"10.1136\/bmjhci-2021-100495","article-title":"Developing, implementing and governing artificial intelligence in medicine: a step-by-step approach to prevent an artificial intelligence winter","volume":"29","author":"van de Sande","journal-title":"BMJ Health Care Inform"},{"key":"R28","doi-asserted-by":"publisher","DOI":"10.2196\/54705","article-title":"AI quality standards in health care: rapid umbrella review","volume":"26","author":"Kuziemsky","journal-title":"J Med Internet Res"},{"key":"R29","unstructured":"Elsworth C Huang K Patterson D et al. Measuring the environmental impact of delivering AI at google scale. arXiv. Preprint posted online on Aug 21, 2025. [doi: 10.48550\/arXiv.2508.15734]"},{"issue":"4","key":"R30","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0302217","article-title":"Does small talk with a medical provider affect ChatGPT\u2019s medical counsel? Performance of ChatGPT on USMLE with and without distractions","volume":"19","author":"Safrai","journal-title":"PLoS One"},{"key":"R31","unstructured":"Cabinet Secretariat, Financial Services Agency, Ministry of Finance, Ministry of Economy, Trade and Industry, Ministry of the Environment. Japan climate transition bonds allocation and impact report for FY2023 issuance\/allocation report for FY2024 issuance. Ministry of Economy, Trade and Industry (METI); 2026. URL: https:\/\/www.meti.go.jp\/policy\/energy_environment\/global_warming\/transition\/climate.transition.bond.allocation.impact.report.eng.pdf [Accessed 06-05-2026]"},{"key":"R32","doi-asserted-by":"publisher","DOI":"10.2196\/69777","article-title":"Trust in AI-supported screening in general practice among urban and rural citizens: cross-sectional study","volume":"14","author":"Wewetzer","journal-title":"JMIR Med Inform"},{"key":"R33","unstructured":"Chen C Yu J Chen S et al. ClinicalBench: can LLMs beat traditional ML models in clinical prediction? arXiv. Preprint posted online on Nov 10, 2024. [doi: 10.48550\/arXiv.2411.06469]"},{"issue":"1","key":"R34","doi-asserted-by":"publisher","DOI":"10.1038\/s41746-025-01653-8","article-title":"Small language models learn enhanced reasoning skills from medical textbooks","volume":"8","author":"Kim","journal-title":"NPJ Digit Med"},{"key":"R35","unstructured":"motokinaru\/LENOHA-medical-dialog. GitHub. URL: https:\/\/github.com\/motokinaru\/LENOHA-medical-dialog [Accessed 02-07-2026]"}],"container-title":["JMIR Medical Informatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e89173","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,31]],"date-time":"2026-07-31T22:24:43Z","timestamp":1785536683000},"score":1,"resource":{"primary":{"URL":"https:\/\/medinform.jmir.org\/2026\/1\/e89173"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,21]]},"references-count":35,"URL":"https:\/\/doi.org\/10.2196\/89173","relation":{},"ISSN":["2291-9694"],"issn-type":[{"value":"2291-9694","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,21]]},"article-number":"v14i4e89173"}}