Demolder Anthony, Nauwynck Maxime, De Pauw Michel, De Buyzere Marc, Duytschaever Mattias, Timmermans Frank, De Pooter Jan
Department of Cardiology, Ghent University Hospital, Ghent, Belgium.
Department of Cardiology, Ghent University Hospital, Ghent, Belgium.
J Electrocardiol. 2024 Mar-Apr;83:71-79. doi: 10.1016/j.jelectrocard.2024.01.008. Epub 2024 Feb 8.
The 12‑lead ECG provides an excellent substrate for artificial intelligence (AI) enabled prediction of various cardiovascular diseases. However, a measure of prediction certainty is lacking.
To assess a novel approach for estimating certainty of AI-ECG predictions.
Two convolutional neural networks (CNN) were developed to predict patient age and sex. Model 1 applied a 5 s sliding time-window, allowing multiple CNN predictions. The consistency of the output values, expressed as interquartile range (IQR), was used to estimate prediction certainty. Model 2 was trained on the full 10s ECG signal, resulting in a single CNN point prediction value. Performance was evaluated on an internal test set and externally validated on the PTB-XL dataset.
Both CNNs were trained on 269,979 standard 12‑lead ECGs (82,477 patients). Model 1 showed higher accuracy for both age and sex prediction (mean absolute error, MAE 6.9 ± 6.3 years vs. 7.7 ± 6.3 years and AUC 0.946 vs. 0.916, respectively, P < 0.001 for both). The IQR of multiple CNN output values allowed to differentiate between high and low accuracy of ECG based predictions (P < 0.001 for both). Among 10% of patients with narrowest IQR, sex prediction accuracy increased from 65.4% to 99.2%, and MAE of age prediction decreased from 9.7 to 4.1 years compared to the 10% with widest IQR. Accuracy and estimation of prediction certainty of model 1 remained true in the external validation dataset.
Sliding window-based approach improves ECG based prediction of age and sex and may aid in addressing the challenge of prediction certainty estimation.
12导联心电图为基于人工智能（AI）预测各种心血管疾病提供了良好的基础。然而，目前缺乏预测确定性的衡量标准。
评估一种估计AI-ECG预测确定性的新方法。
开发了两个卷积神经网络（CNN）来预测患者的年龄和性别。模型1应用5秒的滑动时间窗口，允许进行多次CNN预测。用四分位间距（IQR）表示的输出值一致性用于估计预测确定性。模型2在完整的10秒心电图信号上进行训练，得到单个CNN点预测值。在内部测试集上评估性能，并在PTB-XL数据集上进行外部验证。
两个CNN均在269,979份标准12导联心电图（82,477名患者）上进行训练。模型1在年龄和性别预测方面均显示出更高的准确性（平均绝对误差，MAE分别为6.9±6.3岁和7.7±6.3岁，AUC分别为0.946和0.916，两者P<0.001）。多个CNN输出值的IQR能够区分基于心电图预测的高准确性和低准确性（两者P<0.001）。在IQR最窄的10%患者中，与IQR最宽的10%患者相比，性别预测准确性从65.4%提高到99.2%，年龄预测的MAE从9.7岁降至4.1岁。模型1的准确性和预测确定性估计在外部验证数据集中仍然成立。
基于滑动窗口的方法可改善基于心电图的年龄和性别预测，并可能有助于应对预测确定性估计的挑战。
