EMD与XGBoost组合算法对门诊量预测的研究与分析  

Research and Analysis of EMD and XGBoost Combined Algorithm for the Outpatient Quantity Forecast

作　　者：陈娜 郁晓晨 CHEN Na;YU Xiaochen(Financial Department,Shanghai Sixth People’s Hospital,Shanghai 200233,China)

机构地区：[1]上海市第六人民医院,财务处,上海200233

出　　处：《微型电脑应用》2023年第1期148-151,共4页Microcomputer Applications

摘　　要：利用2016-2019年上海某医院历史门诊量构建数据模型,预测一周日平均和一月日平均门诊量,提出用EMD+XGBoost组合算法在处理日门诊量预测和周门诊量预测。结果表明,该算法优于单XGBoost算法。医院门诊量是医院管理涉及的各种要素中最重要的因素之一,是否能够精确的预测门诊量,对医院的医疗资源配置有着重要的影响。由于医院门诊量是一个非线性时间序列,本文首先利用经验模态分解(EMD)对门诊量序列进行平稳化处理,然后在此基础上增加温度等外部环境因素特征,结合XGBoost算法对门诊量进行预测。实验结果表明,本文提出的EMD+XGBoost组合算法不仅有着较好的预测精度,并且相较于以往的算法,进一步将预测的时间精确到日,效果明显优于单XGBoost算法。Based on the historical outpatient quantity of a hospital in Shanghai from 2016 to 2019, this paper constructs a data model to predict the daily average outpatient quantity of a week and the daily average outpatient quantity of a month. A combined EMD+XGBoost algorithm is proposed to predict the daily outpatient quantity and weekly outpatient quantity. This algorithm is better than the single XGBoost algorithm. Hospital outpatient quantity is one of the most important factors involved in hospital management. Whether we can accurately predict the outpatient quantity has an important impact on the allocation of hospital medical resources. As the hospital outpatient quantity is a nonlinear time series, firstly, the empirical mode decomposition(EMD) is used to stabilize the outpatient quantity series, and then the characteristics of external environmental factors such as temperature are added on this basis, combined with XGBoost algorithm to predict the outpatient quantity. The experimental results show that the proposed EMD+XGBoost combination algorithm not only has better prediction accuracy, but also further improves the prediction time to the day compared with the previous algorithms.

关 键 词：门诊量预测 时间序列 EMD XGBoost 

分 类 号：TP4[自动化与计算机技术]