[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-317032-105":59,"doc-detail-317032-en":135},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":128,"head_meta":130,"extra_data":132,"updated_unix":134},105,"en","automated-assessment-of-osce-physical-exams-using-multimodal-ai","Automated Assessment of OSCE Physical Exams using Multimodal AI","","Evaluation of physical examination skills in medical education is time-consuming and vulnerable to inter-rater variability. This study tests whether a multimodal AI system can grade students’ actual physical techniques, not merely what they verbalize. A retrospective ablation analysis used 300 video-recorded OSCE encounters across six stations, comparing multimodal configurations and input modalities against physician-adjudicated ground truth.",{"@graph":69,"@context":127},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/automated-assessment-of-osce-physical-exams-using-multimodal-ai/317032/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/automated-assessment-of-osce-physical-exams-using-multimodal-ai/317032.png","ImageObject",300,407,{"name":92,"@type":93},"Putri","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-07","2026-09-20",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",12,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119,123],{"name":110,"@type":111,"acceptedAnswer":112},"What problem does the study address in OSCE assessment?","Question",{"text":113,"@type":114},"Traditional OSCE physical-exam grading is resource-intensive and suffers from inter-rater variability. The study targets the difficulty of assessing what students do physically rather than only what they say.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How was the AI system evaluated?",{"text":118,"@type":114},"The authors conducted a retrospective ablation study on 300 video-recorded encounters from six OSCE stations. They compared multimodal AI performance across camera configurations and modality types against physician-adjudicated ground truth using quadratic weighted Cohen’s kappa.",{"name":120,"@type":111,"acceptedAnswer":121},"Which input configuration performed best?",{"text":122,"@type":114},"A synchronized native 3-camera video configuration achieved the highest reliability (κ = 0.830), exceeding standard human evaluators (κ = 0.732). Performance followed a hierarchy: native video > audio-only > transcript-only > visual-only.",{"name":124,"@type":111,"acceptedAnswer":125},"Why did visual-only models underperform despite good detection accuracy?",{"text":126,"@type":114},"Visual-only models produced low agreement (κ ≈ 0.20), reflecting a “visual paradox”: they could identify when an action occurred but could not determine how well it was performed without audio cues.","https://schema.org",{"og:url":83,"og:type":129,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":131,"canonical":83},"index,follow",{"doc_id":133,"site_id":62},317032,1790068926,{"code":4,"msg":5,"data":136},{"doc_id":133,"user_id":137,"nickname":92,"user_avatar":138,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":139,"file_id":140,"file_url":141,"file_type":142,"file_size":143,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":144,"language":145,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":146,"faqs":147,"seo_title":148,"seo_description":67,"update_tm":149,"read_time":150,"preview_page_count":4},962085571259,"https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0","medRxiv preprint doi: [https://doi.org/10.64898/2026.01.09.26343786](https://doi.org/10.64898/2026.01.09.26343786); this version posted January 18, 2026. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  \nAll rights reserved. No reuse allowed without permission.  \nAutomated Assessment of OSCE Physical Exams using Multimodal AI  \nShinyoung Kang1, Michael J. Holcomb1 , Ameer H. Shakur1 , David Hein1 , Huong-Tra Ngo1 , Hunter Schuler1 , Philip C. Jarrett2 , Thomas O. Dalton3 ,  \nAndrew R. Jamieson1*  \n1 Lyda Hill Department of Bioinformatics, University of Texas Southwestern Medical Center. Dallas  \n2 Emergency Medicine, University of Texas Southwestern Medical Center. Dallas  \n3 Department of Internal Medicine, University of Texas Southwestern Medical Center. Dallas  \n* Corresponding author  \nAbstract  \nBackground  \nNOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.  \nmedRxiv preprint doi: [https://doi.org/10.64898/2026.01.09.26343786](https://doi.org/10.64898/2026.01.09.26343786); this version posted January 18, 2026. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  \nAll rights reserved. No reuse allowed without permission.  \nThe assessment of physical examination skills in medical education is resource-intensive and prone to inter-rater variability. While artificial intelligence (AI) has successfully automated the grading of clinical notes and transcripts, evaluating the physical techniques themselves—what students do rather than what they say—remains an unsolved challenge. We evaluated whether a multimodal AI system could assess physical examination skills with expert-level reliability.  \nMethods  \nIn this retrospective ablation study, we analyzed 300 video-recorded encounters from six  \nObjective Structured Clinical Examination (OSCE) stations (cardiovascular, respiratory,  \ngastrointestinal, musculoskeletal, and neurological) . We compared the performance of a  \nmultimodal AI model (Gemini 2.5 Pro) across single-and multi-camera configurations and isolated input modalities (video, audio-only, transcript-only, visual-only) against standard human grading. The primary outcome was agreement with a physician-adjudicated ground-truth  \nreference standard, measured by quadratic weighted Cohen's kappa (κ) .  \nResults  \nThe AI system using a synchronized 3-camera native video configuration achieved significantly higher reliability (κ = 0.830; 95% CI, 0.773–0.880) than the standard human evaluators (κ = 0.732; 95% CI, 0.687–0.776) . Performance followed a strict hierarchy: native video > audio-only> transcript-only > visual-only. Notably, visual-only models failed (κ ≈ 0 .20) despite high detection accuracy, revealing a \"visual paradox\" where models could identify when an action occurred but not how well it was performed without audio cues.  \nConclusions  \nA properly configured multimodal AI system can grade physical examination skills with reliability exceeding that of trained human evaluators. Success requires native processing of synchronized audio-visual streams; transcript-based or visual-only approaches are insufficient for high-stakes assessment. These findings suggest that AI can provide scalable, objective, and valid assessment of clinical skills, overcoming the limitations of traditional human grading.  \nmedRxiv preprint doi: [https://doi.org/10.64898/2026.01.09.26343786](https://doi.org/10.64898/2026.01.09.26343786); this version posted January 18, 2026. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  \nAll rights reserved. No reuse allowed without permission.  \nIntro","cbCaigIBkoUw9r4F","https://ap.wps.com/l/cbCaigIBkoUw9r4F","pdf",1024370,25,"English","# Abstract\n## Background\n## Methods\n## Results\n## Conclusions\n# Introduction","[{\"question\":\"What problem does the study address in OSCE assessment?\",\"answer\":\"Traditional OSCE physical-exam grading is resource-intensive and suffers from inter-rater variability. The study targets the difficulty of assessing what students do physically rather than only what they say.\"},{\"question\":\"How was the AI system evaluated?\",\"answer\":\"The authors conducted a retrospective ablation study on 300 video-recorded encounters from six OSCE stations. They compared multimodal AI performance across camera configurations and modality types against physician-adjudicated ground truth using quadratic weighted Cohen’s kappa.\"},{\"question\":\"Which input configuration performed best?\",\"answer\":\"A synchronized native 3-camera video configuration achieved the highest reliability (κ = 0.830), exceeding standard human evaluators (κ = 0.732). Performance followed a hierarchy: native video \\u003e audio-only \\u003e transcript-only \\u003e visual-only.\"},{\"question\":\"Why did visual-only models underperform despite good detection accuracy?\",\"answer\":\"Visual-only models produced low agreement (κ ≈ 0.20), reflecting a “visual paradox”: they could identify when an action occurred but could not determine how well it was performed without audio cues.\"}]","Automated Assessment of OSCE Physical Exams using Multimodal AI | PDF",1789927984,63]