Taijun Hana, Satoshi Takahashi, Yuki Kawaguchi, S. Oya
{"title":"10048-CO-2 自然语言处理 AI 能否成为脑肿瘤患者的合适 \"顾问\"?聊天技术的临床应用和问题","authors":"Taijun Hana, Satoshi Takahashi, Yuki Kawaguchi, S. Oya","doi":"10.1093/noajnl/vdad141.024","DOIUrl":null,"url":null,"abstract":"Abstract INTRODUCTION Brain tumors affect approximately 20,000 individuals annually in Japan, with some cases requiring urgent treatment. However, the medical knowledge of non-medical individuals (the general public) is often limited, resulting in delays in seeking treatment even when neurological symptoms are evident. Having appropriate advisors who can encourage medical consultations could potentially improve this situation. While Natural Language Processing AI (NLP-AI) is widely used, its full integration into healthcare remains a work in progress. ChatGPT, a prominent NLP-AI, possesses extensive knowledge and is accessible 24/7 online. We explored the potential of leveraging its advanced capabilities and accessibility in the field of brain tumor management. METHODS To evaluate its ability to guide hospital visits effectively, we simulated various symptoms associated with primary brain tumors and consulted ChatGPT(GPT3.5) as patients. We also assessed how ChatGPT handled unnecessary noise information (unrelated symptoms) during symptom consultations. RESULTS ChatGPT consistently recommended hospital visits for all simulated brain tumor patients and provided appropriate advice regarding the urgency and relevant medical department. Notably, even when patients expressed hesitation and refusal hospital visits, ChatGPT persistently persuaded them to visit the hospital despite the user(patient)'s intention. Despite the introduction of noise information, ChatGPT accurately identified and emphasized the importance of relevant symptoms, prompting appropriate medical attention. However, accuracy declined when excessive noise was present. Additionally, ChatGPT generated concise summaries of medical histories for presentation during hospital visits. DISCUSSION While AI is widely utilized in medical imaging, the comprehensive integration of conversational AI in healthcare is still in progress. ChatGPT, despite not being a healthcare professional, holds potential as an \"advisor\" with extensive medical knowledge, accessible for consultations at any time, to support brain tumor patients. However, challenges related to accountability and expanding knowledge capabilities remain to be addressed.","PeriodicalId":19138,"journal":{"name":"Neuro-oncology Advances","volume":" 753","pages":"v6 - v7"},"PeriodicalIF":0.0000,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"10048-CO-2 CAN NATURAL LANGUAGE PROCESSING AI BE AN APPROPRIATE “ADVISER” FOR BRAIN TUMOR PATIENTS? CLINICAL APPLICATION AND ISSUES ON CHATGPT\",\"authors\":\"Taijun Hana, Satoshi Takahashi, Yuki Kawaguchi, S. Oya\",\"doi\":\"10.1093/noajnl/vdad141.024\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Abstract INTRODUCTION Brain tumors affect approximately 20,000 individuals annually in Japan, with some cases requiring urgent treatment. However, the medical knowledge of non-medical individuals (the general public) is often limited, resulting in delays in seeking treatment even when neurological symptoms are evident. Having appropriate advisors who can encourage medical consultations could potentially improve this situation. While Natural Language Processing AI (NLP-AI) is widely used, its full integration into healthcare remains a work in progress. ChatGPT, a prominent NLP-AI, possesses extensive knowledge and is accessible 24/7 online. We explored the potential of leveraging its advanced capabilities and accessibility in the field of brain tumor management. METHODS To evaluate its ability to guide hospital visits effectively, we simulated various symptoms associated with primary brain tumors and consulted ChatGPT(GPT3.5) as patients. We also assessed how ChatGPT handled unnecessary noise information (unrelated symptoms) during symptom consultations. RESULTS ChatGPT consistently recommended hospital visits for all simulated brain tumor patients and provided appropriate advice regarding the urgency and relevant medical department. Notably, even when patients expressed hesitation and refusal hospital visits, ChatGPT persistently persuaded them to visit the hospital despite the user(patient)'s intention. Despite the introduction of noise information, ChatGPT accurately identified and emphasized the importance of relevant symptoms, prompting appropriate medical attention. However, accuracy declined when excessive noise was present. Additionally, ChatGPT generated concise summaries of medical histories for presentation during hospital visits. DISCUSSION While AI is widely utilized in medical imaging, the comprehensive integration of conversational AI in healthcare is still in progress. ChatGPT, despite not being a healthcare professional, holds potential as an \\\"advisor\\\" with extensive medical knowledge, accessible for consultations at any time, to support brain tumor patients. However, challenges related to accountability and expanding knowledge capabilities remain to be addressed.\",\"PeriodicalId\":19138,\"journal\":{\"name\":\"Neuro-oncology Advances\",\"volume\":\" 753\",\"pages\":\"v6 - v7\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2023-12-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Neuro-oncology Advances\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1093/noajnl/vdad141.024\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Neuro-oncology Advances","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1093/noajnl/vdad141.024","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
10048-CO-2 CAN NATURAL LANGUAGE PROCESSING AI BE AN APPROPRIATE “ADVISER” FOR BRAIN TUMOR PATIENTS? CLINICAL APPLICATION AND ISSUES ON CHATGPT
Abstract INTRODUCTION Brain tumors affect approximately 20,000 individuals annually in Japan, with some cases requiring urgent treatment. However, the medical knowledge of non-medical individuals (the general public) is often limited, resulting in delays in seeking treatment even when neurological symptoms are evident. Having appropriate advisors who can encourage medical consultations could potentially improve this situation. While Natural Language Processing AI (NLP-AI) is widely used, its full integration into healthcare remains a work in progress. ChatGPT, a prominent NLP-AI, possesses extensive knowledge and is accessible 24/7 online. We explored the potential of leveraging its advanced capabilities and accessibility in the field of brain tumor management. METHODS To evaluate its ability to guide hospital visits effectively, we simulated various symptoms associated with primary brain tumors and consulted ChatGPT(GPT3.5) as patients. We also assessed how ChatGPT handled unnecessary noise information (unrelated symptoms) during symptom consultations. RESULTS ChatGPT consistently recommended hospital visits for all simulated brain tumor patients and provided appropriate advice regarding the urgency and relevant medical department. Notably, even when patients expressed hesitation and refusal hospital visits, ChatGPT persistently persuaded them to visit the hospital despite the user(patient)'s intention. Despite the introduction of noise information, ChatGPT accurately identified and emphasized the importance of relevant symptoms, prompting appropriate medical attention. However, accuracy declined when excessive noise was present. Additionally, ChatGPT generated concise summaries of medical histories for presentation during hospital visits. DISCUSSION While AI is widely utilized in medical imaging, the comprehensive integration of conversational AI in healthcare is still in progress. ChatGPT, despite not being a healthcare professional, holds potential as an "advisor" with extensive medical knowledge, accessible for consultations at any time, to support brain tumor patients. However, challenges related to accountability and expanding knowledge capabilities remain to be addressed.