基于深度学习与自然语言处理的手术室智慧化管理创新研究  

Research on intelligent management innovation of operating room based on deep learning and natural language processing

作　　者：唐灵逸 郑涛 邵维君 TANG Lingyi;ZHENG Tao;SHAO Weijun(Information Center,Renji Hospital,Shanghai Jiaotong University School of Medicine,Shanghai 200127,China)

机构地区：[1]上海交通大学医学院附属仁济医院信息中心,上海200127

出　　处：《中国数字医学》2023年第8期12-17,共6页China Digital Medicine

摘　　要：在医院手术室日常管理中,医护人员需要根据手术过程和结果选择合适的国际疾病分类(ICD)编码,但具体操作时可能存在错选、乱选等问题。本研究基于自然语言处理与深度学习技术,构建BERT模型、长短期记忆网络模型等,自动纠正手术名称的ICD编码错误并进行ICD编码预测;在此基础上,使用围手术期相关医疗文本数据,基于深度神经网络模型预测手术过程中与术后可能的风险,并提供处理建议,实现手术室的精益化、智慧化运营管理。In the daily management of hospital operating rooms,medical staff need to select the appropriate international classification of diseases(ICD)code in accordance with the surgical procedure and results.Howerer specific operations may encounter problems such as incorrect or arbitrary selection of ICD codes.In this study,BERT model and LSTM(long short-term memory)network model were constructed to automatically correct the ICD coding of surgical names and to predict ICD codes based on natural language processing(NLP)and deep learning(DL)technologies.On this basis,by employing preoperative and intraoperative related medical text data with deep neural network(DNN)models,this study predicts possible risks during and after surgery and provides treatment suggestions to achieve lean and intelligent operation management within the operating room.

关 键 词：国际疾病分类编码 深度学习 自然语言处理 手术室管理 

分 类 号：R197.3[医药卫生—卫生事业管理] R612[医药卫生—公共卫生与预防医学]