【摘要】 随机抽取辽宁省某医院去识别化的急性胰腺炎电子病历,基于支持向量机回归分析方法构建急性胰腺炎患者住院天数预测模型,继而统计分析住院天数的影响因素。所建回归模型预测性能可靠,准确度及泛化性均较好,提出模型建立后的关键是特征变量的选择。更多还原

【Abstract】 By randomly selecting the de-identified electronic medical records of acute pancreatitis in a hospital in Liaoning Province,the paper builds a prediction model of acute pancreatitis patients’ hospital stay based on the regression analysis of Support Vector Machine(SVM) and summarizes factors influencing hospital stay.The prediction of the constructed regression model is reliable with good accuracy and generalization.It is indicated that the key after the model is established lies in the selection of the characteristic variable.更多还原