用大型语言模型检测胃肠道出血,以帮助提高质量和适当的补偿
Neil S Zheng1, Vipina K Keloth2, Kisung You3
1Section of Digestive Diseases, Department of Medicine, Yale School of Medicine, New Haven, Connecticut; Department of Medicine, Brigham and Women's Hospital, Boston, Massachusetts.
Gastroenterology
|September 20, 2024
概括
一个新的大型语言模型 (LLM) 管道准确地识别了电子健康记录中的明显胃肠道出血 (GIB). 这项技术有助于检测复发性出血,并优化GIB患者的报销编码.
科学领域:
- 人工智能的人工智能
- 医疗信息学 医疗信息学
- 胃肠病学 胃肠病学
背景情况:
- 准确识别明显的胃肠道出血 (GIB) 对患者的管理和准确的医疗补偿至关重要.
- 生成型人工智能,特别是大型语言模型 (LLM),为改善电子健康记录 (EHR) 中的临床病症识别提供了潜力.
研究的目的:
- 开发和验证第一个基于LLM的管道,用于在EHR中识别公开的GIB.
- 为了证明管道在两个关键临床应用中的实用性:自动检测反复出血和优化GIB患者的报销编码.
主要方法:
- 一个基于LLM的管道是使用来自1108名急性GIB住院患者的护理笔记来开发的.
- 该管道被用来训练基于EHR的机器学习模型来检测反复出血,以及用于报销编码的算法.
- 模型在来自多家医院的数据上得到验证.
主要成果:
- 该LLM管道在检测黑色素,血和血凝方面表现出高准确性.
- 机器学习模型在检测复发性出血时获得了0.986的AUC,其灵敏度为98.4%,特异性为97.5%.
- 报销编码算法导致了每名患者的平均报销额大幅增加.
结论:
- 一个基于LLM的管道提供了一种强大的方法来检测EHR中的公开GIB.
- 这项技术具有重要的临床应用,用于识别复发性出血并确保GIB病例的适当补偿.
更多相关视频
相关概念视频
01:28Esophageal Varices-II: Clinical Features and Management
50
Esophageal varices often manifest as gastrointestinal bleeding episodes, presenting symptoms like hematemesis (vomiting of blood), hematochezia (passing fresh blood via the rectum), and melena (black, tarry stools). Other signs can include weight loss, anorexia, abdominal discomfort, jaundice, pruritus, altered mental status, and muscle cramps.
In the initial assessment, a thorough review of the patient's medical history is vital to identify risk factors such as liver disease, alcohol...
In the initial assessment, a thorough review of the patient's medical history is vital to identify risk factors such as liver disease, alcohol...
50
01:30Health Information Technology and Healthcare Information System
810
Health Information Technology (HIT)
Health Information Technology, commonly called HIT, integrates advanced information systems and technology in healthcare settings. Its primary functions include:
Health Information Technology, commonly called HIT, integrates advanced information systems and technology in healthcare settings. Its primary functions include:
810