期刊文献+

应用BP神经网络和多重线性回归模型预测颅内出血患者住院天数 被引量:3

Comparison of BP neural network versus multiple linear regression model in predicting the hospitalized stay in patients with intracerebral hemorrhage
下载PDF
导出
摘要 目的研究BP神经网络模型与多重线性回归模型对颅内出血患者住院天数的预测性能。方法回顾中国人民解放军第309医院2011-01~2014-05入院患者诊断为颅内出血的病例资料。应用SPSS 19.0实现BP神经网络模型和多重线性回归模型对住院天数进行预测分析。应用R、R2、调整R2、标准误差和平均相对误差作为评价指标对两模型的预测结果进行比较,得出各模型的预测效果。结果 BP神经网络模型预测住院天数的R2=0.484,调整R2=0.483;多重线性回归模型预测住院天数的R2=0.467,调整R2=0.460。二者预测结果相比较,BP神经网络模型具有优势(t=4.099,P〈0.001)。BP神经网络和多重线性回归预测住院天数的标准误差分别为7.188和7.389;平均相对误差分别为36.810%和37.101%。结论本研究中BP神经网络模型的预测性能优于多重线性回归模型,对颅内出血临床路径制定有参考价值。 Objective To compare the value of back propagation( BP) neural network versus multiple linear regression for predicting the hospitalized stay in patients with intracerebral hemorrhage( ICH). Methods Clinical data of patients with ICH were collected in309 th Hospital of Chinese PLA from January 2011 to May 2014. SPSS19. 0 software was applied to analyze the length of hospitalization predicted by BP neural network and multiple linear regression. R,R2,adjusted R2,standard error and average relative error were used in evaluation of the two models. Results Predictive values for the hospitalized stay were R^2= 0. 484 and adjusted R2= 0. 483 by BP neural network,and R2= 0. 467 and adjusted R2= 0. 460 for multiple linear regression. The predicting performance of BP neural network was better than that of multiple linear regression( t = 4. 099,P 〈0. 001). The standard error was 7. 188 for BP neural network and7. 389 for multiple linear regressions,and the average relative error was 36. 810% and 37. 101%,respectively. Conclusion The predictive performance of BP neural network is better than that of multiple linear regressions,and it has a reference value for the clinical pathway of intracerebral hemorrhage.
出处 《山西医科大学学报》 CAS 2015年第10期1007-1010,共4页 Journal of Shanxi Medical University
基金 解放军第309医院科研基金资助项目(2014MS-009)
关键词 颅内出血 BP神经网络 多重线性回归 数学模型 intracerebral hemorrhage BP neural network multiple linear regression mathematical model
  • 相关文献

参考文献13

二级参考文献86

共引文献134

同被引文献28

引证文献3

二级引证文献6

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部