Background and objective

Sudden cardiac death (SCD) is an unexpected loss in functioning of the heart. It generally occurs within 1 h of onset of symptoms. No reliable method for early detection of SCD is available. Therefore, the development of a non-invasive method for risk identification remains a topic of utmost interest. This paper presents a novel approach for detection of the risk of SCD by performing a comparative analysis of heart rate variability (HRV) in normal subjects as well as patients with coronary artery disease and heart failure.

Methods

HRV of four subject groups, normal sinus rhythm, coronary artery disease, congestive heart failure, and SCD, has been analyzed. The analysis was performed by using nonlinear techniques and time-frequency representation obtained by generalized S-transform. The extracted features were examined for their clinical significance by using the Kruskal–Wallis one-way analysis of variance and multiple comparisons. Eventually, classification was performed using support vector machines and decision tree classifiers to separate the individuals at risk of SCD.

Results

The performance of the proposed methodology has been evaluated using PhysioNet open-access databases. Statistical analysis shows that HRV in SCD group differs significantly from other groups. For classification, an accuracy of 91.67% was achieved with 83.33% sensitivity, 94.64% specificity, and 84.75% precision.

Conclusion

The experimental results obtained by analyzing retrospective data seem promising. However, the methodology needs to be tested on larger databases to generalize the findings. Prospective studies on its clinical usefulness may help in developing a concrete diagnostic technique.

中文翻译：

---

通过比较正常和异常心脏状况下心率变异性的心律失常检测

背景和目标

心脏猝死（SCD）是心脏功能的意外损失。它通常在症状发作后1小时内发生。没有可靠的早期检测SCD的方法。因此，开发用于风险识别的非侵入性方法仍然是最受关注的话题。本文通过对正常受试者以及患有冠心病和心力衰竭的患者进行心率变异性（HRV）的比较分析，提出了一种检测SCD风险的新颖方法。

方法

分析了四个受试者组的HRV，正常窦性心律，冠状动脉疾病，充血性心力衰竭和SCD。使用非线性技术进行分析，并通过广义S变换获得时频表示。通过使用Kruskal-Wallis单向方差分析和多重比较检查提取的特征的临床意义。最终，使用支持向量机和决策树分类器进行分类以分离有SCD风险的个体。

结果

已使用PhysioNet开放访问数据库评估了所提出方法的性能。统计分析表明，SCD组的HRV与其他组明显不同。对于分类，准确度达到91.67％，灵敏度为83.33％，特异性为94.64％，准确度为84.75％。

结论

通过分析回顾性数据获得的实验结果似乎很有希望。但是，该方法需要在较大的数据库上进行测试以概括研究结果。关于其临床实用性的前瞻性研究可能有助于开发具体的诊断技术。