Jiang Min, Ke Jian, Fang Ming-Hao, Huang Su-Fang, Li Yuan-Yuan
Department of Emergency Medicine, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, 430030, China.
Curr Med Sci. 2023 Oct;43(5):961-969. doi: 10.1007/s11596-023-2768-8. Epub 2023 Jul 14.
It is difficult to predict fulminant myocarditis at an early stage in the emergency department. The objective of this study was to construct and validate a simple prediction model for the early identification of fulminant myocarditis.
A total of 61 patients with fulminant myocarditis and 160 patients with acute myocarditis were enrolled in the training and internal validation cohorts. LASSO regression and multivariate logistic regression were selected to develop the prediction model. The selection of the model was based on overall performance and simplicity. A nomogram based on the optimal model was built, and its clinical usefulness was evaluated by decision curve analysis. The predictive model was further validated in an external validation group.
The resulting prediction model was based on 4 factors: systolic blood pressure, troponin I, left ventricular ejection fraction, and ventricular wall motion abnormality. The Brier scores of the final model were 0.078 in the training data set and 0.061 in the internal testing data set, respectively. The C-indexes of the training data set and the testing data set were 0.952 and 0.968, respectively. Decision curve analysis showed that the nomogram model developed based on the 4 predictors above had a positive net benefit for predicting probability thresholds. In the external validation cohort, the model also showed good performance (Brier score=0.007, and C-index=0.989).
We developed and validated an early prediction model consisting of 4 clinical factors (systolic blood pressure, troponin I, left ventricular ejection fraction, and ventricular wall motion abnormality) to identify potential fulminant myocarditis patients in the emergency department.
在急诊科很难在早期预测暴发性心肌炎。本研究的目的是构建并验证一个用于早期识别暴发性心肌炎的简单预测模型。
共有61例暴发性心肌炎患者和160例急性心肌炎患者被纳入训练和内部验证队列。选择套索回归和多变量逻辑回归来建立预测模型。模型的选择基于整体性能和简单性。构建了基于最优模型的列线图，并通过决策曲线分析评估其临床实用性。在外部验证组中进一步验证了该预测模型。
最终的预测模型基于4个因素：收缩压、肌钙蛋白I、左心室射血分数和室壁运动异常。最终模型在训练数据集和内部测试数据集中的Brier评分分别为0.078和0.061。训练数据集和测试数据集的C指数分别为0.952和0.968。决策曲线分析表明，基于上述4个预测因子建立的列线图模型在预测概率阈值方面具有正的净效益。在外部验证队列中，该模型也表现出良好的性能（Brier评分为0.007，C指数为0.989）。
我们开发并验证了一个由4个临床因素（收缩压、肌钙蛋白I、左心室射血分数和室壁运动异常）组成的早期预测模型，以在急诊科识别潜在的暴发性心肌炎患者。
