目的评价3种机器学习算法模型对脑梗死患者是否伴有颅内动脉狭窄的预测性能,为缺血性卒中诊治策略提供参考依据。方法回顾性收集自2013年1月至2019年9月喀什地区第一人民医院神经内科收治的缺血性脑卒中患者临床资料。通过特征选择筛选...目的评价3种机器学习算法模型对脑梗死患者是否伴有颅内动脉狭窄的预测性能,为缺血性卒中诊治策略提供参考依据。方法回顾性收集自2013年1月至2019年9月喀什地区第一人民医院神经内科收治的缺血性脑卒中患者临床资料。通过特征选择筛选出脑梗死伴颅内动脉狭窄的相关因素作为预测因子,基于随机森林、决策树和神经网络3种机器学习算法建立预测模型。利用受试者工作特征曲线下面积(area under ROC,AUC)、灵敏度、准确率等指标评价3种模型对脑梗死患者是否伴有颅内动脉狭窄的预测性能。结果研究分析了2365例脑梗死患者的74种特征,通过特征选择选出23个特征作为预测因子纳入建模。决策树模型、神经网络模型、随机森林模型的AUC值分别为0.78±0.11、0.85±0.12、0.89±0.10;灵敏度依次为0.92±0.05、0.91±0.06、0.88±0.10;准确率依次为0.79±0.12、0.77±0.13、0.85±0.13。结论随机森林模型对脑梗死患者是否伴有颅内动脉狭窄的预测性能最佳,在无条件进行血管成像检查时可应用于脑梗死是否伴有颅内动脉狭窄的预测,具有一定的临床应用价值。展开更多
A simple but illustrative survey is given on various approaches of computational intelligence with their features, applications and the mathematical tools involved, among which the simulated annealing, neural networks...A simple but illustrative survey is given on various approaches of computational intelligence with their features, applications and the mathematical tools involved, among which the simulated annealing, neural networks, genetic and evolutionary programming, self-organizing learning and adapting algorithms, hidden Markov models are recommended intensively. The common mathematical features of various computational intelligence algorithms are exploited.Finally, two common principles of concessive strategies implicated in many computational intelligence algorithms are discussed.展开更多
文摘目的评价3种机器学习算法模型对脑梗死患者是否伴有颅内动脉狭窄的预测性能,为缺血性卒中诊治策略提供参考依据。方法回顾性收集自2013年1月至2019年9月喀什地区第一人民医院神经内科收治的缺血性脑卒中患者临床资料。通过特征选择筛选出脑梗死伴颅内动脉狭窄的相关因素作为预测因子,基于随机森林、决策树和神经网络3种机器学习算法建立预测模型。利用受试者工作特征曲线下面积(area under ROC,AUC)、灵敏度、准确率等指标评价3种模型对脑梗死患者是否伴有颅内动脉狭窄的预测性能。结果研究分析了2365例脑梗死患者的74种特征,通过特征选择选出23个特征作为预测因子纳入建模。决策树模型、神经网络模型、随机森林模型的AUC值分别为0.78±0.11、0.85±0.12、0.89±0.10;灵敏度依次为0.92±0.05、0.91±0.06、0.88±0.10;准确率依次为0.79±0.12、0.77±0.13、0.85±0.13。结论随机森林模型对脑梗死患者是否伴有颅内动脉狭窄的预测性能最佳,在无条件进行血管成像检查时可应用于脑梗死是否伴有颅内动脉狭窄的预测,具有一定的临床应用价值。
文摘A simple but illustrative survey is given on various approaches of computational intelligence with their features, applications and the mathematical tools involved, among which the simulated annealing, neural networks, genetic and evolutionary programming, self-organizing learning and adapting algorithms, hidden Markov models are recommended intensively. The common mathematical features of various computational intelligence algorithms are exploited.Finally, two common principles of concessive strategies implicated in many computational intelligence algorithms are discussed.