论文部分内容阅读
目的探讨和评价随机森林法与Logistic回归用于预测预约挂号失约影响因素的可行性与准确性。方法以新疆某三甲医院的预约挂号患者为研究对象,随机抽取预约挂号患者400份,回顾性分析预约患者的特性及失约行为,利用随机森林法与Logistic回归分别建立预约挂号失约影响因素的模型。结果对于预约患者失约影响因素预测分析,随机森林回归明显优于Logistic回归方法,随机森林回归对失约因素误判率(20.5%)明显低于Logistic回归(22%);用随机森林方法显示失约的影响因素的顺序是:年龄、具体时间、短信提醒、工作状况。结论完善短信提醒功能,降低失约率;预测易失约人群,合理分配号源,避免医疗资源的浪费。
Objective To explore and evaluate the feasibility and accuracy of the factors influencing the default of the registered appointment by the random forest method and Logistic regression. Methods A total of 400 registered patients were randomly selected from a top three hospital in Xinjiang to analyze the characteristics of appointment patients and the default behavior. The randomized forest method and Logistic regression were used to establish the model of influencing factors of appointment loss. Results Prediction analysis showed that random forest regression was superior to Logistic regression method. The misjudgment rate (20.5%) of random forest regression was significantly lower than that of Logistic regression (22%). Stochastic forest method showed that the missed The order of influencing factors is: age, specific time, SMS reminder, working condition. Conclusion Improve the SMS reminding function and reduce the failure rate; predict the vulnerable population, rationally allocate the source and avoid the waste of medical resources.