[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-367834-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-367834-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","robotic-assistemt-an-emt-chatbot-stanford-cs224n-natural-language-processing-with-deep-learning","Robotic AssistEMT: An EMT Chatbot - Stanford CS224N Natural Language Processing with Deep Learning","","Development of AssistEMT, an Emergency Medical Technician (EMT) chatbot, to support emergency medical examinations through empathetic conversation and guidance aligned with NREMT standards. The work addresses limited medical-depth knowledge and lack of empathy in general chatbots by adding EMT-specific datasets and training/fine-tuning GPT-3.5 and GPT-4. Retrieval-Augmented Generation (RAG) enables timely updates to medical knowledge. Prompt engineering ablations further improve assistant-like demeanor, user trust, and performance in EMT knowledge exams and on-scene assessment scenarios.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/robotic-assistemt-an-emt-chatbot-stanford-cs224n-natural-language-processing-with-deep-learning/367834/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/robotic-assistemt-an-emt-chatbot-stanford-cs224n-natural-language-processing-with-deep-learning/367834.png","ImageObject",300,407,{"name":42,"@type":43},"404brain","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-11-08","2026-09-23",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",18,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What is AssistEMT and what tasks does it support?","Question",{"text":63,"@type":64},"AssistEMT is an EMT chatbot designed to conduct medical interviews and assessments, helping EMTs evaluate patients in emergency settings and synthesize information for diagnosis and transport decisions.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How do fine-tuning and RAG contribute to the system?",{"text":68,"@type":64},"Fine-tuning uses EMT-specific datasets to improve domain knowledge performance. RAG enables real-time updates of medical knowledge so the chatbot stays current and maintains accuracy and ethical compliance.",{"name":70,"@type":61,"acceptedAnswer":71},"Why does the project conclude GPT-4 performs best?",{"text":72,"@type":64},"The project finds GPT-4’s extended context and inference capabilities are most suitable for EMT chatbot tasks, improving emergency medical assessments compared with alternatives.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},367834,1790194894,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":118,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":144,"read_time":104,"preview_page_count":33,"preview_urls":145},1099529839291,"https://ap-avatar.wpscdn.com/davatar_6f874abed73319feea01a86fa6f0fab8","Robotic AssistEMT: An EMT Chatbot  \nStanford CS224N Custom Project  \nAanika Atluri  \nDepartment of Computer Science Stanford University [aatluri@stanford.edu](aatluri@stanford.edu)[ ](aatluri@stanford.edu)Examples of more authors  \nSarah Barragan  \nDepartment of Computer Science Stanford University [sabarrag@stanford.edu](sabarrag@stanford.edu)  \nAnusheh Chaudry  \nDepartment of Computer Science  \nStanford University  \n[anusheh@stanford.edu](anusheh@stanford.edu)  \nAbstract  \nWe present the development of an Emergency Medical Technician (EMT) chatbot, AssistEMT, designed to serve as a conversational assistant for emergency medical examinations and to interface with robots capable of performing necessary physical assessments. Despite the critical role of AI in various domains, its integration into emergency medicine remains underexplored due to the highly uncertain and critical nature of the field. We aim to address this gap by developing an empathetic EMT chatbot capable of accurately assessing patients in accordance with the National Registry of Emergency Medical Technicians (NREMT) guidelines and synthesizing information to provide diagnoses and transport decisions. Our contributions include the creation of two specialized datasets centered around EMT-specific medical knowledge. These datasets were then used to fine-tune a GPT-3.5 model ChatGPT (2024b) and to implement Retrieval-Augmented Generation (RAG) Lewis et al.  \n(2021) with GPT-4 ChatGPT (2024a) . Our major findings reveal that GPT-4’s extensive context and inference capabilities make it the most well-suited for the tasks of an EMT chatbot. The incorporation of RAG further augments the chatbot’s functionality by enabling real-time updates of medical knowledge, ensuring the chatbot remains updated with the latest medical discoveries. Our prompt engineering ablation studies have also resulted in a chatbot that operates with an empathetic demeanor akin to a human assistant, enhancing user experience and trust.  \n1 Key Information to include  \n• Mentor: Kenneth Salisbury, Sunny Singh (External Mentors) & Rashon Poole  \n• External Collaborators (if you have any): Anjana Balachandar & Varsha Naga (Trained EMTs)  \n• Sharing project: No  \n• Group Contributions: Anusheh worked on the GPT-3.5 fine tuned model and creating our own datasets, Aanika worked on the GPT-4 with prompt engineering model as well asablation studies and automating dataset creation, and Sarah worked on the GPT-4 with RAG model as well contributing to the creation of the datasets. We all contributed to the evaluation of our models and to the final report.  \nStanford CS224N Natural Language Processing with Deep Learning  \n2 Introduction  \nEmergency Medical Technicians (EMT) face critical challenges in being able to rapidly assess and respond to patient needs during emergencies. However there has been a recent emergence of AI chatbots transforming healthcare by processing vast data and detecting patterns beyond human capability, potentially alleviating the healthcare system’s load and allowing medical professionals to focus on complex tasks. However, the quality of these chatbots is limited by their training data. While they can provide general health information and manage routine tasks, they struggle with queries requiring deep medical knowledge, critical thinking, and years of clinical experience. Additionally, AI lacks the empathy and intuition of human medical professionals. Thus, the integration of chatbots with medical professionals has evolved into a collaborative approach, where chatbots supplement physicians in complex decision-making and empathetic patient care.  \nOur project aims to bridge this gap through developing a Robotic EMT chatbot interface that leverages NLP techniques to conduct medical interviews and assessments assisting an EMT. To address persisting challenges in the field, we present the development of an EMT (Emergency Medical Technician) chatbot using fine-tuning and Retrieval-Augm","cbCaifh8nEcJL4SV","https://ap.wps.com/l/cbCaifh8nEcJL4SV","pdf",564787,"English","# Abstract\n# Key Information to include\n# Introduction\n# Related Work","[{\"question\":\"What is AssistEMT and what tasks does it support?\",\"answer\":\"AssistEMT is an EMT chatbot designed to conduct medical interviews and assessments, helping EMTs evaluate patients in emergency settings and synthesize information for diagnosis and transport decisions.\"},{\"question\":\"How do fine-tuning and RAG contribute to the system?\",\"answer\":\"Fine-tuning uses EMT-specific datasets to improve domain knowledge performance. RAG enables real-time updates of medical knowledge so the chatbot stays current and maintains accuracy and ethical compliance.\"},{\"question\":\"Why does the project conclude GPT-4 performs best?\",\"answer\":\"The project finds GPT-4’s extended context and inference capabilities are most suitable for EMT chatbot tasks, improving emergency medical assessments compared with alternatives.\"}]","Robotic AssistEMT: An EMT Chatbot - Stanford CS224N Natural Language Processing with Deep Learning | PDF",1790172905,[146,150,154,158],{"page_no":22,"url":147,"width":148,"height":149},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/269cc628201589e3fcfb969617fb8c46.webp",983,1272,{"page_no":26,"url":151,"width":152,"height":153},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/9f487ada70aac7e3f0642f7b330887e9.webp",982,1270,{"page_no":30,"url":155,"width":156,"height":157},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/d241d610a9aa81bd7d92ccba41b786ad.webp",981,1269,{"page_no":33,"url":159,"width":160,"height":161},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/88862bf96173da080ede0624c235bf28.webp",980,1268]