《中国卫生质量管理》[ISSN:1006-7515/CN:CN 61-1283/R]
卷:
第24卷
期数:
2017年04期
页码:
15-17
栏目:
医疗质量
出版日期:
2017-07-28
- 作者:
-
绳慧峰 刘晴 许苹 陈春林
-
解放军第105医院
- 关键词:
-
人工神经网络; 径向基函数; 医疗风险; 预测; 应用
- Keywords:
-
Artificial Neural Network; Radial Basis Function; Medical Risk; Prediction; Application
- DOI:
-
10.13912/j.cnki.chqm.2017.24.4.07
- 摘要:
-
目的建立基于人工神经网络的医疗风险预测模型，为有效预防医疗风险提供参考。方法运用SPSS 21.0统计软件，以入院方式、住院天数等变量为输入神经元，以医疗纠纷分组为输出神经元，用RBF（径向基函数）建立神经网络模型，预测医疗风险的发生，评估各因素对医疗风险的作用。结果训练样本和测试样本的预测准确率分别为83.7%和84.2%，医疗风险影响因素重要性排序前6位分别是住院费用（100.0%）、住院天数（78.2%）、四周内手术次数（61.4%）、感染（60.5%）、伤口愈合不良（54.0%）和手术并发症（47.8%）等。结论运用RBF（径向基函数）神经网络对医疗风险进行预测，不受样本分布特点及数据类型的影响，适用性较好。
- Abstract:
-
Objective To establish prediction model of medical risks based on artificial neural network, provide reference for effective prevention of medical risks. MethodsWith the admission form, length of stay and other variables as the input neurons, medical disputes group as the output neuron, using radial basis function (RBF) to establish neural network model, the occurrence of medical risk was predicted, and the factors on the role of health risk was evaluated. ResultsThe prediction accuracies of the training sample and test sample were 83.7% and 84.2%, respectively. The top six importance of medical risk factors were hospitalization expenses (100.0%), length of stay (78.2%), four weeks’ operation times (61.4%), infection (60.5%), poor wound healing (54.0%) and complications (47.8%). ConclusionIt is not affected by sample distribution features and data types to use RBF neural network to forecast the medical risks and has a good applicability.
更新日期/Last Update:
2017-07-28
