In the article by van Westrhenen,¹ an inconsistency between the most recent results and those provided in the article as a miscalculation in the data analysis for one of the subjects was found by the authors. As a result, all results of the included cases were re-checked, and it was determined misleading outcomes were provided for six of the 22 subjects reported because a discordant standard was used. Results for the remaining sixteen are unchanged.

Once the proper standard is used, a lower number of true positives and a higher number of false alarms are seen. The number of false alarms is 87 rather than 81. Consequently, the overall sensitivity of the video algorithm for the detection of convulsive seizures is 78% rather than the 94% reported. The detection performance for non-convulsive seizures was also affected: overall sensitivity for detecting hyperkinetic seizures 73% rather than the 100% reported.

A corrected version of Figure 1 follows.

We apologize for these errors.

在 van Westrhenen 的文章¹中，作者发现最新结果与文章中提供的结果之间存在不一致，这是作者对其中一个受试者的数据分析中的错误计算。结果，对纳入病例的所有结果进行了重新检查，并确定报告的 22 名受试者中有 6 名提供了误导性结果，因为使用了不一致的标准。其余十六个的结果没有变化。

一旦使用了正确的标准，就会看到更少的真阳性和更多的误报。误报的数量是 87 次而不是 81 次。因此，用于检测惊厥发作的视频算法的总体灵敏度为 78%，而不是报告的 94%。非惊厥性癫痫发作的检测性能也受到影响：检测多动性癫痫发作的总体灵敏度为 73%，而不是报告的 100%。

图 1 的修正版本如下。

对于这些错误，我们深表歉意。