基于电子病历数据的疾病预测方法  

Disease prediction method based on electronic medical record data

作　　者：张博 刘经纬[2] Zhang Bo;Liu Jingwei(984 Hospital of Joint Logistic Support Force,Beijing 100094,China;North China Institute of Computing Technology,Beijing 100083,China)

机构地区：[1]联勤保障部队第九八四医院,北京100094 [2]华北计算技术研究所,北京100083

出　　处：《现代科学仪器》2024年第1期161-166,共6页Modern Scientific Instruments

摘　　要：为进一步加强临床诊断的准确性,提高疾病预测的精准度。因此提出基于电子病历数据的妊娠糖尿病预测方法。首先,在建模之前,通过对样本的基线精度进行评估,验证学习算法的正确性,并在此基础上对样本进行均衡。消除因数据类别不均衡导致的模型预测结果过度偏倚。在本试验中,采用均方误差以验证该方法的精度。通过训练集构建逻辑回归模型,在预测模型中引入测试集的数据。Logistic回归的F1值和AUC值分别为0.809、0.881和0.825,与不使用该特征时相比增加了约12%。结果表明,电子病历数据驱动可以有效提高妊娠糖尿病预测的准确性。To further enhance the accuracy of clinical diagnosis and improve the precision of disease prediction.Therefore,a prediction method for gestational diabetes based on electronic medical record data is proposed.Firstly,the correctness of the learning algorithm is verified by evaluating the baseline accuracy of the samples before modelling,and the samples are balanced on this basis.Excessive bias in the model prediction results due to unbalanced data categories is eliminated.In this experiment,mean square error is used to validate the accuracy of the method.A logistic regression model was constructed from the training set and data from the test set was introduced into the prediction model.the F1 and AUC values for the logistic regression were 0.809,0.881 and 0.825 respectively,an increase of approximately 12%compared to when the feature was not used.The results suggest that electronic medical record data-driven can be effective in improving the accuracy of gestational diabetes prediction.

关 键 词：电子病历 数据 疾病预测 临床诊断 

分 类 号：R197.323[医药卫生—卫生事业管理] R18[医药卫生—公共卫生与预防医学]