基于灰色预测与人工神经网络模型的大型综合公立医院住院收入预测研究

Research on the forecast of income in large comprehensive public hospital based on grey prediction and artificial neural network model

目的为了全面分析医院业务收入的影响因素,准确预测医院的收入,从而为医院编制预算、制订长远的发展战略、进行医院基础设施建设规划,促进医疗卫生事业的发展提供参考。方法利用山东省某区域医疗中心以及山东省统计年鉴的相关数据,通过Lasso变量选择方法选择公立医院住院收入的影响因素,分析了医院自身发展指标以及国民经济与社会发展状况相关指标对住院收入的影响。采用灰色预测和人工神经网络模型对大型公立医院住院收入进行预测。结果经变量选择,纳入住院收入预测模型的指标有:地区生产总值、年末人口数、入院人数、住院手术量和床位周转次数,其中地区生产总值、年末人口数、入院人数和住院手术量的增多,有助于医院住院收入的提升;床位周转次数的降低,有利于提高大型公立医院的住院收入。经灰色预测,各预测指标的预测相对精度均在90%以上,经人工神经网络预测得2016年某医院住院收入的预测值为245 003万元。结论灰色预测和人工神经网络组合模型,较好地实现了公立医院住院收入的预测,预测结果优于单纯灰色预测的预测结果。灰色预测和人工神经网络组合预测模型综合考虑了被预测变量的多个影响因素,因而比单纯的灰色预测法更加稳健。

Objective To comprehensively analyze the influencing factors of hospital business income,and accurately forecast hospital revenues,thus,to provide references for the hospital to compile budget,formulate long-term development strategy,carry out infrastructure construction plan,and promote the development of medical and health undertakings.Methods The related data of a regional medical center and statistical yearbook of shandong province was used,and the influencing factors of in-patient income in public hospitals were selected by using the Lasso variable selection method.The influences of the hospital's own development index and the related indexes of national economy and social development on the income of hospital were analyzed,and grey prediction and artificial neural network model were used to predict the inpatient income of large public hospitals.Results According to the variable selection,the indicators included in the forecast model of hospital income were regional GDP,number of people at the end of the year,number of hospital admissions,number of hospital operations and number of bed turnover.The increase of regional GDP,number of people at the end of the year,number of hospital admissions,and number of hospital operations helped improve the hospital income.The decrease of bed turnover was beneficial to the increase of the income of large public hospitals.According to the grey prediction,the relative accuracy of each prediction index was above 90%,and the predicted value of hospital inpatient income in 2016 was 245 003 by the artificial neural network.Conclusion The combined model of grey prediction and artificial neural network can well predict the income of inpatient in public hospitals.This model takes into account several influencing factors of the predicted variables,therefore,compared with the simple grey prediction method,it is more robust.