Sudden cardiac arrest (SCA) is a lethal cardiac arrhythmia that poses a serious threat to human life and health. However, clinical records of sudden cardiac death (SCD) electrocardiogram (ECG) data are extremely limited. This paper proposes an early prediction and classification algorithm for SCA based on deep transfer learning. With limited ECG data, it extracts heart rate variability features before the onset of SCA and utilizes a lightweight convolutional neural network model for pre-training and fine-tuning in two stages of deep transfer learning. This achieves early classification, recognition and prediction of high-risk ECG signals for SCA by neural network models. Based on 16 788 30-second heart rate feature segments from 20 SCA patients and 18 sinus rhythm patients in the international publicly available ECG database, the algorithm performance evaluation through ten-fold cross-validation shows that the average accuracy (Acc), sensitivity (Sen), and specificity (Spe) for predicting the onset of SCA in the 30 minutes prior to the event are 91.79%, 87.00%, and 96.63%, respectively. The average estimation accuracy for different patients reaches 96.58%. Compared to traditional machine learning algorithms reported in existing literatures, the method proposed in this paper helps address the requirement of large training datasets for deep learning models and enables early and accurate detection and identification of high-risk ECG signs before the onset of SCA.
心脏骤停（SCA）是一种致命性心律失常，会对人体生命健康造成严重威胁。基于目前临床记录的心脏猝死（SCD）心电图（ECG）数据极其有限，本文提出了一种基于深度迁移学习的心脏骤停早期预估及分类算法。本文算法在有限的ECG数据训练下，通过提取心脏骤停发作前的心率变异性特征，并送入轻量级卷积神经网络模型进行预训练和微调训练两个阶段的深度迁移学习，实现神经网络模型对心脏骤停高危ECG信号的早期分类识别和预估。基于国际公开ECG数据库中20个心脏猝死患者和18个窦性心律患者的16 788条30 s心率特征片段，本文采用十折交叉量化验证的算法性能评估实验结果显示，对心脏骤停发作前30 min预测的平均准确度（Acc）、灵敏度（Sen）和特异度（Spe）分别为91.79%、87.00%和96.63%；而对不同患者的平均预估准确度达到96.58%。相较于已报道的传统机器学习算法，本文方法不仅有助于解决深度学习模型对大量训练数据的要求，而且能够更加早期、准确地检测和识别心脏骤停发作前的高危ECG征兆。.

Track and trigger systems (TTSs) based on vital signs are implemented in hospitals worldwide to identify patients with clinical deterioration. TTSs may provide prognostic information but do not actively include clinical assessment, and their impact on severe adverse events remain uncertain. The demand for prospective, multicentre studies to demonstrate the effectiveness of TTSs has grown the last decade. Individual Early Warning Score (I-EWS) is a newly developed TTS with an aggregated score based on vital signs that can be adjusted according to the clinical assessment of the patient. The objective is to compare I-EWS with the existing National Early Warning Score (NEWS) algorithm regarding clinical outcomes and use of resources.
In a prospective, multicentre, cluster-randomised, crossover, non-inferiority study. Eight hospitals are randomised to use either NEWS in combination with the Capital Region of Denmark NEWS Override System (CROS) or implement I-EWS for 6.5 months, followed by a crossover. Based on their clinical assessment, the nursing staff can adjust the aggregated score with a maximum of -4 or +6 points. We expect to include 150 000 unique patients. The primary endpoint is all-cause mortality at 30 days. Coprimary endpoint is the average number of times per day a patient is NEWS/I-EWS-scored, and secondary outcomes are all-cause mortality at 48 hours and at 7 days as well as length of stay.
The study was presented for the Regional Ethics committee who decided that no formal approval was needed according to Danish law (J.no. 1701733). The I-EWS study is a large prospective, randomised multicentre study that investigates the effect of integrating a clinical assessment performed by the nursing staff in a TTS, in a head-to-head comparison with the internationally used NEWS with the opportunity to use CROS.
NCT03690128.