College of Biomedical Engineering and Instrument Science, Zhejiang University, Key Lab for Biomedical Engineering of Ministry of Education, Zheda Road, Hangzhou, China.
Department of Cardiology, Chinese PLA General Hospital, Beijing, China.
BMC Med Inform Decis Mak. 2019 Jan 9;19(1):5. doi: 10.1186/s12911-018-0730-7.
Main adverse cardiac events (MACE) are essentially composite endpoints for assessing safety and efficacy of treatment processes of acute coronary syndrome (ACS) patients. Timely prediction of MACE is highly valuable for improving the effects of ACS treatments. Most existing tools are specific to predict MACE by mainly using static patient features and neglecting dynamic treatment information during learning.
We address this challenge by developing a deep learning-based approach to utilize a large volume of heterogeneous electronic health record (EHR) for predicting MACE after ACS. Specifically, we obtain the deep representation of dynamic treatment features from EHR data, using the bidirectional recurrent neural network. And then, the extracted latent representation of treatment features can be utilized to predict whether a patient occurs MACE in his or her hospitalization.
We validate the effectiveness of our approach on a clinical dataset containing 2930 ACS patient samples with 232 static feature types and 2194 dynamic feature types. The performance of our best model for predicting MACE after ACS remains robust and reaches 0.713 and 0.764 in terms of AUC and Accuracy, respectively, and has over 11.9% (1.2%) and 1.9% (7.5%) performance gain of AUC (Accuracy) in comparison with both logistic regression and a boosted resampling model presented in our previous work, respectively. The results are statistically significant.
We hypothesize that our proposed model adapted to leverage dynamic treatment information in EHR data appears to boost the performance of MACE prediction for ACS, and can readily meet the demand clinical prediction of other diseases, from a large volume of EHR in an open-ended fashion.
主要心脏不良事件（MACE）是评估急性冠状动脉综合征（ACS）患者治疗过程安全性和疗效的综合终点。及时预测 MACE 对于改善 ACS 治疗效果具有重要价值。大多数现有的工具主要使用静态患者特征来预测 MACE，而在学习过程中忽略了动态治疗信息。
我们通过开发一种基于深度学习的方法来解决这一挑战，该方法利用大量异构电子健康记录（EHR）来预测 ACS 后 MACE。具体来说，我们使用双向递归神经网络从 EHR 数据中获取动态治疗特征的深度表示。然后，可以利用提取的治疗特征的潜在表示来预测患者在住院期间是否发生 MACE。
我们在包含 2930 名 ACS 患者样本的临床数据集上验证了我们方法的有效性，该数据集包含 232 种静态特征类型和 2194 种动态特征类型。我们用于预测 ACS 后 MACE 的最佳模型的性能仍然稳健，在 AUC 和准确性方面分别达到 0.713 和 0.764，与我们之前工作中提出的逻辑回归和增强重采样模型相比，AUC（准确性）分别提高了 11.9%（1.2%）和 1.9%（7.5%）。结果具有统计学意义。
我们假设我们提出的模型能够适应利用 EHR 数据中的动态治疗信息，从而提高 ACS 中 MACE 预测的性能，并且可以从大量的 EHR 中以开放的方式满足其他疾病的临床预测需求。
