心脏骤停后缺氧性昏迷的预后预测:自动化脑电图分析方法的比较
该研究回顾性分析荷兰五家教学医院ICU中871例心脏骤停后3天内记录的头颅连续脑电图,比较逻辑回归(2个定量特征)、随机森林(9个特征)和卷积神经网络深度学习模型对6个月结局(CPC 1-2为良好,CPC 3-5为不良)的预测性能。两中心数据用于训练和五折交叉验证(n=663),另三中心数据用于外部验证(n=208)。外部测试集上,深度学习网络在24小时预测不良结局的敏感性为54%(95%CI 44-64%)、假阳性率0%(95%CI 0-2%),显著高于逻辑回归(敏感性33%,FPR 0%)和随机森林(敏感性13%,FPR 0%)(p<0.05);12小时预测良好结局的敏感性为78%(95%CI 52-100%)、FPR 12%(95%CI 0-24%)。深度学习结果受伪差、白噪声和通道平坦化影响最小。
要点
- 外部验证集中,深度学习模型24小时预测不良结局的敏感性54%(95%CI 44-64%)、FPR 0%(95%CI 0-2%),高于逻辑回归33%和随机森林13%(FPR均0%,p<0.05)。
- 12小时预测良好结局时,深度学习模型敏感性78%(95%CI 52-100%)、FPR 12%(95%CI 0-24%),逻辑回归敏感性83%(95%CI 83-83%)、FPR 3%(95%CI 3-3%),两者均显著高于随机森林(敏感性1%,FPR 0%,p<0.05)。
- 深度学习网络对伪差、人工添加白噪声和随机平坦化通道的稳健性最好,结果受影响最小。
- 研究为回顾性分析,训练集来自两中心(n=663),外部验证集来自另三中心(n=208),模型输出为良好结局概率。
适用人群与场景
- 心脏骤停后入住ICU的昏迷患者,连续脑电图记录至骤停后3天内。
- 结局评估为6个月时CPC评分,1-2分定义为良好结局,3-5分定义为不良结局。
诊断标准与关键阈值
- 结局二分类标准:6个月脑功能表现分类(Cerebral Performance Category,CPC)1-2分为良好,3-5分为不良。
- 预测模型输出为良好结局概率,通过受试者工作特征分析和预测值计算评估性能。
研究设计与关键结果
- 研究类型为回顾性分析,共纳入871例连续脑电图,来自荷兰五家教学医院ICU;训练集来自两中心(n=663),外部验证集来自另三中心(n=208)。
- 比较三种模型:逻辑回归(2个定量特征)、随机森林(9个特征)和基于卷积神经网络的深度学习模型。
- 外部测试集24小时预测不良结局:深度学习敏感性54%(95%CI 44-64%)、FPR 0%(95%CI 0-2%);逻辑回归敏感性33%、FPR 0%;随机森林敏感性13%、FPR 0%;深度学习显著高于后两者(p<0.05)。
- 外部测试集12小时预测良好结局:深度学习敏感性78%(95%CI 52-100%)、FPR 12%(95%CI 0-24%);逻辑回归敏感性83%(95%CI 83-83%)、FPR 3%(95%CI 3-3%);两者均显著高于随机森林(敏感性1%、FPR 0%)(p<0.05)。
- 稳健性评估显示,深度学习网络结果受伪差、人工添加白噪声及随机平坦化通道的影响最小。
证据类型、等级与局限
- 证据类型:回顾性多中心研究(五家教学医院),训练集两中心(n=663),外部验证集三中心(n=208)。
- 局限:原文未报告局限性。
对床旁实践的改变
- 对于心脏骤停后昏迷患者,基于连续脑电图的深度学习模型在外部验证中表现出比逻辑回归和随机森林更高的预测不良结局敏感性(54% vs 33% vs 13%,FPR均为0%),且对伪差更稳健,可能有助于床旁预后判断。
- 深度学习模型在12小时预测良好结局的敏感性为78%、FPR 12%,逻辑回归敏感性83%、FPR 3%,两者均优于随机森林(敏感性1%、FPR 0%)。
- 作者认为深度学习模型在基于脑电图的可靠、稳健的结局预测方面优于逻辑回归和随机森林模型。
提炼说明
- 本条依据论文摘要整理;摘要未涉及「处置流程与药物剂量」「红旗征与不能漏的情况」「指南或共识出处」,故未列出这些栏目;全文请点「阅读原文」。
英文原文(摘要)· 点击展开
BACKGROUND: To compare three computer-assisted quantitative electroencephalography (EEG) prediction models for the outcome prediction of comatose patients after cardiac arrest regarding predictive performance and robustness to artifacts.
METHODS: A total of 871 continuous EEGs recorded up to 3 days after cardiac arrest in intensive care units of five teaching hospitals in the Netherlands were retrospectively analyzed. Outcome at 6 months was dichotomized as "good" (Cerebral Performance Category 1-2) or "poor" (Cerebral Performance Category 3-5). Three prediction models were implemented: a logistic regression model using two quantitative features, a random forest model with nine features, and a deep learning model based on a convolutional neural network. Data from two centers were used for training and fivefold cross-validation (n = 663), and data from three other centers were used for external validation (n = 208). Model output was the probability of good outcome. Predictive performances were evaluated by using receiver operating characteristic analysis and the calculation of predictive values. Robustness to artifacts was evaluated by using an artifact rejection algorithm, manually added noise, and randomly flattened channels in the EEG.
RESULTS: The deep learning network showed the best overall predictive performance. On the external test set, poor outcome could be predicted by the deep learning network at 24 h with a sensitivity of 54% (95% confidence interval [CI] 44-64%) at a false positive rate (FPR) of 0% (95% CI 0-2%), significantly higher than the logistic regression (sensitivity 33%, FPR 0%) and random forest models (sensitivity 13%, FPR, 0%) (p < 0.05). Good outcome at 12 h could be predicted by the deep learning network with a sensitivity of 78% (95% CI 52-100%) at a FPR of 12% (95% CI 0-24%) and by the logistic regression model with a sensitivity of 83% (95% CI 83-83%) at a FPR of 3% (95% CI 3-3%), both significantly higher than the random forest model (sensitivity 1%, FPR 0%) (p < 0.05). The results of the deep learning network were the least affected by the presence of artifacts, added white noise, and flat EEG channels.
CONCLUSIONS: A deep learning model outperformed logistic regression and random forest models for reliable, robust, EEG-based outcome prediction of comatose patients after cardiac arrest.
参考来源
- Pham SDT, Keijzer HM, Ruijter BJ, et al. Outcome Prediction of Postanoxic Coma: A Comparison of Automated Electroencephalography Analysis Methods. Neurocrit Care. 2022. PMID: 35233717. doi:10.1007/s12028-022-01449-8
供执业医师抢救与诊疗参考,不替代临床判断;剂量请结合体重、肝肾功能和本院药品规格复核,以指南原文为准。