Detection of Patients with Congenital and Often Concealed Long-QT Syndrome by Novel Deep Learning Models

Abstract

Introduction: The long-QT syndrome (LQTS) is the most common ion channelopathy, typically presenting with a prolonged QT interval and clinical symptoms such as syncope or sudden cardiac death. Patients may present with a concealed phenotype making the diagnosis challenging. Correctly diagnosing at-risk patients is pivotal to starting early preventive treatment. Objective: Identification of congenital and often concealed LQTS by utilizing novel deep learning network architectures, which are specifically designed for multichannel time series and therefore particularly suitable for ECG data. Design and Results: A retrospective artificial intelligence (AI)-based analysis was performed using a 12-lead ECG of genetically confirmed LQTS (n = 124), including 41 patients with a concealed LQTS (33%), and validated against a control cohort (n = 161 of patients) without known LQTS or without QT-prolonging drug treatment but any other cardiovascular disease. The performance of a fully convolutional network (FCN) used in prior studies was compared with a different, novel convolutional neural network model (XceptionTime). We found that the XceptionTime model was able to achieve a higher balanced accuracy score (91.8%) than the associated FCN metric (83.6%), indicating improved prediction possibilities of novel AI architectures. The predictive accuracy prevailed independently of age and QTc parameters. Conclusions: In this study, the XceptionTime model outperformed the FCN model for LQTS patients with even better results than in prior studies. Even when a patient cohort with cardiovascular comorbidities is used. AI-based ECG analysis is a promising step for correct LQTS patient identification, especially if common diagnostic measures might be misleading.

A MATLAB algorithm to automatically estimate the QT interval and other ECG parameters and validation using a machine learning approach in congenital long-QT syndrome

“…The authors found that using QTc alone, AUC was 0.82, whereas using all 12-leads, 10 s each, as an input to a deep neural network algorithm led to an AUC of 0.9, suggesting that additional parameters in the ECG signal could help to better identify LQTS patients. This result has been replicated (Doldi et al, 2022) who showed using a novel convolutional neural network model resulting in a high accuracy rate (91.8%) in identification of concealed LQTS. In another study, (Aufiero et al, 2022) used neural networks on ECG data to show that genotype positive LQTS patients can be accurately identified based on the ECG data only.…”

“…The latter can reduce the risk of sudden cardiac death (SCD) in affected individuals and related, potentially affected, family members. Correct and precise detection of LQTS in the surface ECG can be, however, challenging and resemble a random finding, e.g., when associated with non-specific symptoms, in the setting of intermittent, heart-rate dependent ECG changes, including so-called concealed QTc prolongation (Doldi et al, 2022; Sugrue et al, 2016), and due to a genetic complexity (Ingles et al, 2020). Furthermore, the lack of routine ECG screening (Van Der Werf et al, 2010) and reliance on manual measurements or automatic estimations by ECG analysis software, leads to unrecognized or misjudged diagnoses of LQT syndrome.…”

Deep Learning–Augmented ECG Analysis for Screening and Genotype Prediction of Congenital Long QT Syndrome

“…A growing body of work has shown that ML approaches are effective in the analysis of ECGs. In terms of congenital LQTS detection, ML algorithms have been developed, including an XceptionTime model to detect LQTS based on a small cohort of patients with high accuracy scores . Some CNN-based models have been developed in single-center studies to predict a clinical or genetic diagnosis of LQTS, and another has been developed with randomly selected outpatient control ECGs and validated in an external center with high accuracy .…”

“…Recently, different machine learning (ML) approaches, including support vector machines and convolutional neural networks (CNNs), have been applied to the detection of LQTS on ECGs . Building on existing approaches, we aimed to develop a CNN model that could detect LQTS on baseline ECGs.…”

“…5 Recently, different machine learning (ML) approaches, including support vector machines and convolutional neural networks (CNNs), have been applied to the detection of LQTS on ECGs. [6][7][8][9] Building on existing approaches, we aimed to develop a CNN model that could detect LQTS on baseline ECGs. We tested our model on a large and ethnically diverse group of patients who were referred to quaternary centers for screening of suspected LQTS based on phenotype or family history.…”