基于支持向量机的呼吸器肺炎检测算法

Detection of Ventilator-Associated Pneumonia in Intensive Care Unit Based on Support Vector Machine

针对重症监护室病人接受呼吸器插管治疗后,肺炎感染状况无法得到实时监控和诊断的问题,提出了一种基于支持向量机模型的呼吸器肺炎检测方案。使用电子鼻装置采集未注射抗生素病人的呼出气体作为实验数据,建立了支持向量机肺炎检测模型,利用交叉验证和受试者工作特征曲线对模型的稳定性和准确性进行分析。实验结果表明,支持向量机模型对呼吸器肺炎感染检测是非常稳定、有效的,为医师早期用药提供科学有效的参考。

Concerning that patients are not detected in real time for whether infected pneumonia after using the ventilator in intensive care units（ ICU）,a new solution for ventilator-associated pneumonia（ VAP） diagnose was proposed. Firstly,in order to prevent the variation of pneumonia strains,the breath gas of non-injected antibiotics was collected by Cyranose-320 electronic nose as the experimental data,the support vector machine（ SVM）method was used to build pneumonia recognition model; secondly,six fold cross-validation method was applied to evaluate the stability of SVM model; finally,for the purpose of evaluating the accuracy of the model,the area under the receiver operating characteristic curve（ ROC） was calculated. The results show that SVM model have high recognition of pneumonia,the accuracy（ ACC） is 0.8789 ± 0.0359,sensitivity（ SEN） is 0.9202 ± 0.0556,and positive predictive value（ PPV） is 0. 8513 ± 0. 0409. Besides,the area under the ROC curve（ AUC） is0. 9419±0.0350,shown that the SVM model is stable and a good classifier. This study aims to predict patients whether infected with pneumonia through SVM model,providing a scientific and effective reference for physician to perform early diagnosis.