基于深度学习模型的智能化科室导诊  

Intelligent department guidance based on deep learning model

作　　者：顾君杰 王蓓[1] 李晓禹 邹俊忠[1] GU Jun-jie;WANG Bei;LI Xiao-yu;ZOU Jun-zhong(School of Information Science and Engineering,East China University of Science and Technology,Shanghai 200237,China;Research and Development Department,Qingying Medical Technology(Shenzhen)Limited Company,Shenzhen 518083,China)

机构地区：[1]华东理工大学信息科学与工程学院,上海200237 [2]清影医疗科技(深圳)有限公司研发部,广东深圳518083

出　　处：《计算机工程与设计》2024年第1期153-158,共6页Computer Engineering and Design

基　　金：国家自然科学基金面上基金项目(61773164)。

摘　　要：为减轻科室导诊人员的工作负荷,对智能化科室导诊的实现方法进行研究。区别于现有的导诊方式,提出一种少参数轻量化的多级科室导诊模型。结合ALBERT预训练解决现有算法参数量过大的问题,并关联多个相关科室,建立ALBERT预训练与Bi-GRU结合的多标签分类模型。通过在互联网医院问诊数据集上的测试,与单科室分类模型对比,验证了该多科室分类模型的预测结果具备可靠性和有效性,能够较好辅助科室导诊工作。To reduce the workload of traditional department guidance,the intelligent department guidance method was studied.Different from the existing intelligent guidance methods,a multi-department guidance model with fewer parameters was proposed.ALBERT pre-training was implemented to solve the problem of large number of parameters in existing algorithms.A multi-department classification model combining ALBERT pre-training and BI-GRU was constructed.Based on the evaluation using collected consultation data set and the comparison with single-department classification model,it is verified that the presented multi-department model is more reliable and effective,it can better assist the department guidance work.

关 键 词：科室导诊 多标签 文本预训练 双向门控循环单元 文本分类 深度学习 自然语言处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]