基于深度学习的医疗辅助诊断数据分析方法  

Data analysis method of medical assistant diagnosis based on deep learning

作　　者：程顺达[1] 祝婕[1] 夏芳[1] CHENG Shunda;ZHU Jie;XIA Fang(Information Center,Hebei Hospital of Traditional Chinese Medicine,Shijiazhuang 050000,China)

机构地区：[1]河北省中医院信息中心,河北石家庄050000

出　　处：《电子设计工程》2022年第7期72-76,共5页Electronic Design Engineering

基　　金：2021年度河北省医学科学研究课题计划(20210100);2020年河北省中医药管理局科研计划项目(KTY2020104)。

摘　　要：针对海量医疗诊断数据缺乏高效且准确的分析方法,文中提出了一种基于深度学习的医疗辅助诊断数据处理方法。该方法对医疗诊断数据进行预处理,包括归一化处理与主成分分析,以得到医疗数据特征;利用部署多层感知器的神经网络分析已完成预处理的数据,从而实现诊断分类,为医生提供智能化的辅助诊断;基于Python软件平台对所提方法展开实验测试。结果表明,所提方法的损失值与平均准确率分别为53和85%,均优于其他对比方法,且具有良好的工程应用价值。In view of the lack of efficient and accurate analysis method for massive medical diagnosis data,a method of medical auxiliary diagnosis data analysis based on deep learning is proposed.The medical diagnosis data is preprocessed,including normalization and principal component analysis,and the characteristics of medical data are obtained.The neural network with multi-layer perceptron is used to analyze the preprocessed data,so as to realize diagnosis classification and provide intelligent auxiliary diagnosis for doctors.The proposed method is tested based on Python software platform,and the results show that the loss value and average accuracy rate of the proposed method are 53 and 85%respectively,which are better than other comparison methods and have good engineering application value.

关 键 词：深度学习 医疗辅助诊断 数据分析 主成分分析 多层感知器神经网络 

分 类 号：TN99[电子电信—信号与信息处理] TP18[电子电信—信息与通信工程]