基于深度信念网络的医院门诊量预测

发布时间：2018-02-13 21:53

  本文关键词： 深度信念网络 门诊量预测 数据特征 逻辑回归　出处：《计算机科学》2016年S2期 　论文类型：期刊论文

【摘要】：有效的医院门诊量预测是现代医院对医疗资源实现智能化管理的重要前提之一。现有的医院门诊量预测方法大多针对的是单一的数据集,缺少对数据的充分挖掘和深入分析。为此,提出一种基于深度信念网络的医院门诊量预测方法,用深度信念网络对医院各科室的门诊量数据进行无监督学习,完成对门诊量数据的特征提取,挖掘各科室门诊量数据间的相互关系,在网络的顶层叠加一个逻辑回归层并将提取出的数据特征作为输入来预测各科室未来的门诊量。仿真实验结果表明,基于深度学习的预测模型可以得到较高的门诊量预测精度,是一种可行且有效的预测方法。
[Abstract]:Effective outpatient volume prediction is one of the important premises for modern hospital to realize intelligent management of medical resources. Most of the existing methods of outpatient volume prediction are aimed at a single data set. For this reason, a method of outpatient quantity prediction based on deep belief network is proposed, which can be used to study the outpatient data of each department without supervision. Complete the feature extraction of the outpatient data, mining the relationship between the outpatient data of each department, A logical regression layer is superimposed on the top of the network, and the extracted data feature is used as input to predict the future outpatient volume of each department. The simulation results show that the prediction model based on in-depth learning can obtain higher prediction accuracy of outpatient volume. It is a feasible and effective prediction method.
【作者单位】： 浙江工业大学计算机科学与技术学院;
【基金】：国家自然科学基金(61374152)资助
【分类号】：R197.3;TP183
，

本文编号：1509183