Department of Molecular and Clinical Medicine, Institute of Medicine, University of Gothenburg, Wallenberg Laboratory, Blå stråket 5, Sahlgrenska University Hospital, 413 45 Gothenburg, Sweden; Department of Cardiology, Sahlgrenska University Hospital, Blå stråket 5, 413 45 Gothenburg, Sweden; The Swedish Registry for Cardiopulmonary Resuscitation, Medicinaregatan 18G, 413 90 Gothenburg, Sweden.
Department of Anesthesiology and Intensive Care, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Blå stråket 5, 413 45 Gothenburg, Sweden; Department of Anesthesiology and Intensive Care, Sahlgrenska University Hospital, Göteborgsvägen 31, 431 30 Mölndal, Sweden.
Resuscitation. 2024 Sep;202:110359. doi: 10.1016/j.resuscitation.2024.110359. Epub 2024 Aug 12.
Out-of-hospital cardiac arrest (OHCA) is a critical condition with low survival rates. In patients with a return of spontaneous circulation, brain injury is a leading cause of death. In this study, we propose an interpretable machine learning approach for predicting neurologic outcome after OHCA, using information available at the time of hospital admission.
The study population were 55 615 OHCA cases registered in the Swedish Cardiopulmonary Resuscitation Registry between 2010 and 2020. The dataset was split to training and validation sets (for model development) and test set (for evaluation of the final model). We used an XGBoost algorithm with stratified, repeated 10-fold cross-validation along with Optuna framework for hyperparameters tuning. The final model was trained on 10 features selected based on the importance scores and evaluated on the test set in terms of discrimination, calibration and bias-variance tradeoff. We used SHapley Additive exPlanations to address the 'black-box' model and align with eXplainable artificial intelligence.
The final model achieved: area under the receiver operating characteristic value 0.964 (95% confidence interval (CI) [0.960-0.968]), sensitivity 0.606 (95% CI [0.573-0.634]), specificity 0.975 (95% CI [0.972-0.978]), positive predictive value (PPV) 0.664 (95% CI [0.625-0.696]), negative predictive value (NPV) 0.969 (95% CI [0.966-0.972]), macro F1 0.803 (95% CI [0.788-0.816]), and showed a very good calibration. SHAP features with the highest impact on the model's output were:'ROSC on arrival to hospital', 'Initial rhythm asystole' and 'Conscious on arrival to hospital'.
The XGBoost machine learning model with 10 features available at the time of hospital admission showed good performance for predicting neurologic outcome after OHCA, with no apparent signs of overfitting.
院外心脏骤停（OHCA）是一种生存率较低的危急病症。在自主循环恢复的患者中，脑损伤是导致死亡的主要原因。本研究提出了一种可解释的机器学习方法，用于预测 OHCA 后的神经功能结局，使用入院时的可用信息。
本研究人群为 2010 年至 2020 年期间在瑞典心肺复苏登记处登记的 55615 例 OHCA 病例。数据集分为训练集和验证集（用于模型开发）以及测试集（用于评估最终模型）。我们使用 XGBoost 算法，结合分层、重复 10 折交叉验证和 Optuna 框架进行超参数调整。最终模型基于重要性得分选择的 10 个特征进行训练，并在测试集上评估模型的判别能力、校准能力和偏差-方差权衡。我们使用 SHapley Additive exPlanations 解决“黑箱”模型问题，并与可解释人工智能保持一致。
最终模型的表现为：接收者操作特征曲线下面积 0.964（95%置信区间[0.960-0.968]），敏感度 0.606（95%置信区间[0.573-0.634]），特异性 0.975（95%置信区间[0.972-0.978]），阳性预测值（PPV）0.664（95%置信区间[0.625-0.696]），阴性预测值（NPV）0.969（95%置信区间[0.966-0.972]），宏 F1 0.803（95%置信区间[0.788-0.816]），且具有很好的校准能力。对模型输出影响最大的 SHAP 特征是：“入院时 ROSC”“初始节律为停搏”和“入院时意识清醒”。
基于入院时可用的 10 个特征的 XGBoost 机器学习模型对 OHCA 后神经功能结局的预测具有良好的性能，且没有明显的过拟合迹象。
