Improving long QT syndrome diagnosis by a polynomial-based T-wave morphology characterization

Application of artificial intelligence in the diagnosis and treatment of cardiac arrhythmia

The rapid growth in computational power, sensor technology, and wearable devices has provided a solid foundation for all aspects of cardiac arrhythmia care. Artificial intelligence (AI) has been instrumental in bringing about significant changes in the prevention, risk assessment, diagnosis, and treatment of arrhythmia. This review examines the current state of AI in the diagnosis and treatment of atrial fibrillation, supraventricular arrhythmia, ventricular arrhythmia, hereditary channelopathies, and cardiac pacing. Furthermore, ChatGPT, which has gained attention recently, is addressed in this paper along with its potential applications in the field of arrhythmia. Additionally, the accuracy of arrhythmia diagnosis can be improved by identifying electrode misplacement or erroneous swapping of electrode position using AI. Remote monitoring has expanded greatly due to the emergence of contactless monitoring technology as wearable devices continue to develop and flourish. Parallel advances in AI computing power, ChatGPT, availability of large data sets, and more have greatly expanded applications in arrhythmia diagnosis, risk assessment, and treatment. More precise algorithms based on big data, personalized risk assessment, telemedicine and mobile health, smart hardware and wearables, and the exploration of rare or complex types of arrhythmia are the future direction.

Applying Artificial Intelligence for Phenotyping of Inherited Arrhythmia Syndromes

Inherited arrhythmia disorders account for a significant proportion of sudden cardiac death, particularly among young individuals. Recent advances in our understanding of these syndromes have improved patient diagnosis and care, yet certain clinical gaps remain, particularly within case ascertainment, access to genetic testing, and risk stratification. Artificial intelligence (AI), specifically machine learning and its subset deep learning, present promising solutions to these challenges. The capacity of AI to process vast amounts of patient data and identify disease patterns differentiates them from traditional methods, which are time- and resource-intensive. To date, AI models have shown immense potential in condition detection (including asymptomatic/concealed disease) and genotype and phenotype identification, exceeding expert cardiologists in these tasks. Additionally, they have exhibited applicability for general population screening, improving case ascertainment in a set of conditions that are often asymptomatic such as left ventricular dysfunction. Third, models have shown the ability to improve testing protocols; through model identification of disease and genotype, specific clinical testing (eg, drug challenges or further diagnostic imaging) can be avoided, reducing health care expenses, speeding diagnosis, and possibly allowing for more incremental or targeted genetic testing approaches. These significant benefits warrant continued investigation of AI, particularly regarding the development and implementation of clinically applicable screening tools. In this review we summarize key developments in AI, including studies in long QT syndrome, Brugada syndrome, hypertrophic cardiomyopathy, and arrhythmogenic cardiomyopathies, and provide direction for effective future AI implementation in clinical practice.

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

Importance

Congenital long QT syndrome (LQTS) is associated with syncope, ventricular arrhythmias, and sudden death. Half of patients with LQTS have a normal or borderline-normal QT interval despite LQTS often being detected by QT prolongation on resting electrocardiography (ECG).

Objective

To develop a deep learning-based neural network for identification of LQTS and differentiation of genotypes (LQTS1 and LQTS2) using 12-lead ECG.

Design, Setting, and Participants

This diagnostic accuracy study used ECGs from patients with suspected inherited arrhythmia enrolled in the Hearts in Rhythm Organization Registry (HiRO) from August 2012 to December 2021. The internal dataset was derived at 2 sites and an external validation dataset at 4 sites within the HiRO Registry; an additional cross-sectional validation dataset was from the Montreal Heart Institute. The cohort with LQTS included probands and relatives with pathogenic or likely pathogenic variants in KCNQ1 or KCNH2 genes with normal or prolonged corrected QT (QTc) intervals.

Exposures

Convolutional neural network (CNN) discrimination between LQTS1, LQTS2, and negative genetic test results.

Main Outcomes and Measures

The main outcomes were area under the curve (AUC), F1 scores, and sensitivity for detecting LQTS and differentiating genotypes using a CNN method compared with QTc-based detection.

Results

A total of 4521 ECGs from 990 patients (mean [SD] age, 42 [18] years; 589 [59.5%] female) were analyzed. External validation within the national registry (101 patients) demonstrated the CNN's high diagnostic capacity for LQTS detection (AUC, 0.93; 95% CI, 0.89-0.96) and genotype differentiation (AUC, 0.91; 95% CI, 0.86-0.96). This surpassed expert-measured QTc intervals in detecting LQTS (F1 score, 0.84 [95% CI, 0.78-0.90] vs 0.22 [95% CI, 0.13-0.31]; sensitivity, 0.90 [95% CI, 0.86-0.94] vs 0.36 [95% CI, 0.23-0.47]), including in patients with normal or borderline QTc intervals (F1 score, 0.70 [95% CI, 0.40-1.00]; sensitivity, 0.78 [95% CI, 0.53-0.95]). In further validation in a cross-sectional cohort (406 patients) of high-risk patients and genotype-negative controls, the CNN detected LQTS with an AUC of 0.81 (95% CI, 0.80-0.85), which was better than QTc interval-based detection (AUC, 0.74; 95% CI, 0.69-0.78).

Conclusions and Relevance

The deep learning model improved detection of congenital LQTS from resting ECGs and allowed for differentiation between the 2 most common genetic subtypes. Broader validation over an unselected general population may support application of this model to patients with suspected LQTS.

The Heart of Rett Syndrome: A Quantitative Analysis of Cardiac Repolarization

Background

Rett syndrome (RTT) is a developmental encephalopathy disorder that is associated with a high incidence of sudden death presumably from cardiorespiratory etiologies. Electrocardiogram (ECG) abnormalities, such as prolonged heart-rate corrected QT (QTc) interval, are markers of cardiac repolarization and are associated with potentially lethal ventricular arrhythmias. This study investigates the cardiac repolarization characteristics of RTT patients, including QTc and T-wave morphology characteristics.

Methods

A retrospective quantitative analysis on 110 RTT patients and 124 age and sex-matched healthy controls was conducted.

Results

RTT patients had longer QTc, more abnormal T-wave morphology, and greater heterogeneity of cardiac repolarization parameters compared to controls. Even RTT patients without prolonged QTc had more abnormal ECG and T-wave characteristics than controls. Among RTT patients, MECP2 patients had prolonged QTc compared to CDKL5 and FOXG1 patients. A subset of five RTT patients who died had normal QTc, but more abnormal T-wave morphology than the remaining RTT patients.

Conclusions

Cardiac repolarization abnormalities are present in RTT patients, even without long QTc. T-wave morphology is related to RTT genotype and may be predictive of mortality. These findings could be used to help the management and monitoring of RTT patients.

Provocation testing in congenital long QT syndrome: A practical guide

Congenital long QT syndrome (LQTS) is a hereditary cardiac channelopathy with an estimated prevalence of 1 in 2500. A prolonged resting QT interval corrected for heart rate (QTc interval) remains a key diagnostic component; however, the QTc value may be normal in up to 40% of patients with genotype-positive LQTS and borderline in a further 30%. Provocation of QTc prolongation and T-wave changes may be pivotal to unmasking the diagnosis and useful in predicting genotype. LQTS provocation testing involves assessment of repolarization during and after exercise, in response to changes in heart rate or autonomic tone, with patients with LQTS exhibiting a maladaptive repolarization response. We review the utility and strengths and limitations of 4 forms of provocation testing-stand-up test, exercise stress test, epinephrine challenge, and mental stress test-in diagnosing LQTS and provide some practical guidance for performing provocation testing. Ultimately, exercise testing, when feasible, is the most useful form of provocation testing when considering diagnostic sensitivity and specificity.

Toward advanced diagnosis and management of inherited arrhythmia syndromes: Harnessing the capabilities of artificial intelligence and machine learning

The use of advanced computational technologies, such as artificial intelligence (AI), is now exerting a significant influence on various aspects of life, including health care and science. AI has garnered remarkable public notice with the release of deep learning models that can model anything from artwork to academic papers with minimal human intervention. Machine learning, a method that uses algorithms to extract information from raw data and represent it in a model, and deep learning, a method that uses multiple layers to progressively extract higher-level features from the raw input with minimal human intervention, are increasingly leveraged to tackle problems in the health sector, including utilization for clinical decision support in cardiovascular medicine. Inherited arrhythmia syndromes are a clinical domain where multiple unanswered questions remain despite unprecedented progress over the past 2 decades with the introduction of large panel genetic testing and the first steps in precision medicine. In particular, AI tools can help address gaps in clinical diagnosis by identifying individuals with concealed or transient phenotypes; enhance risk stratification by elevating recognition of underlying risk burden beyond widely recognized risk factors; improve prediction of response to therapy, and further prognostication. In this contemporary review, we provide a summary of the AI models developed to solve challenges in inherited arrhythmia syndromes and also outline gaps that can be filled with the development of intelligent AI models.

Secular trends of health care resource utilization and costs between Brugada syndrome and congenital long QT syndrome: A territory-wide study

BACKGROUND

Health care resource utilization (HCRU) and costs are important metrics of health care burden, but they have rarely been explored in the setting of cardiac ion channelopathies.

HYPOTHESIS

This study tested the hypothesis that attendance-related HCRUs and costs differed between patients with Brugada syndrome (BrS) and congenital long QT syndrome (LQTS).

METHODS

This was a retrospective cohort study of consecutive BrS and LQTS patients at public hospitals or clinics in Hong Kong, China. HCRUs and costs (in USD) for Accident and Emergency (A&E), inpatient, general outpatient and specialist outpatient attendances were analyzed between 2001 and 2019 at the cohort level. Comparisons were made using incidence rate ratios (IRRs [95% confidence intervals]).

RESULTS

Over the 19-year period, 516 BrS (median age of initial presentation: 51 [interquartile range: 38-61] years, 92% male) and 134 LQTS (median age of initial presentation: 21 [9-44] years, 32% male) patients were included. Compared to LQTS patients, BrS patients had lower total costs (2 008 126 [2 007 622-2 008 629] vs. 2 343 864 [2 342 828-2 344 900]; IRR: 0.857 [0.855-0.858]), higher costs for A&E attendances (83 113 [83 048-83 177] vs. 70 604 [70 487-70 721]; IRR: 1.177 [1.165-1.189]) and general outpatient services (2,176 [2,166-2,187] vs. 921 [908-935]; IRR: 2.363 [2.187-2.552]), but lower costs for inpatient stay (1 391 624 [1 391 359-1 391 889] vs. 1 713 742 [1 713 166-1 714 319]; IRR: 0.812 [0.810-0.814]) and lower costs for specialist outpatient services (531 213 [531 049-531 376] vs. 558 597 [558268-558926]; IRR: 0.951 [0.947-0.9550]).

CONCLUSIONS

Overall, BrS patients consume 14% less health care resources compared to LQTS patients in terms of attendance costs. BrS patients require more A&E and general outpatient services, but less inpatient and specialist outpatient services than LQTS patients.

The diagnostic value of electrocardiogram-based machine learning in long QT syndrome: a systematic review and meta-analysis

Introduction

To perform a meta-analysis to discover the performance of ML algorithms in identifying Congenital long QT syndrome (LQTS).

Methods

The searched databases included Cochrane, EMBASE, Web of Science, and PubMed. Our study considered all English-language studies that reported the detection of LQTS using ML algorithms. Quality was assessed using QUADAS-2 and QUADAS-AI tools. The bivariate mixed effects models were used in our study. Based on genotype data for LQTS, we performed a subgroup analysis.

Results

Out of 536 studies, 8 met all inclusion criteria. The pooled area under the receiving operating curve (SAUROC) for detecting LQTS was 0.95 (95% CI: 0.31-1.00); sensitivity was 0.87 (95% CI: 0.83-0.90), and specificity was 0.91 (95% CI: 0.88-0.93). Additionally, diagnostic odd ratio (DOR) was 65 (95% CI: 39-109). The positive likelihood ratio (PLR) was 9.3 (95% CI: 7.0-12.3) and the negative likelihood ratio (NLR) was 0.14 (95% CI: 0.11-0.20), with very low heterogeneity (I²= 16%).

Discussion

We found that machine learning can be used to detect features of rare cardiovascular disease like LQTS, thus increasing our understanding of intelligent interpretation of ECG. To improve ML performance in the classification of LQTS subtypes, further research is required.

Systematic Review Registration

identifier PROSPERO CRD42022360122.

Precision medicine for long QT syndrome: patient-specific iPSCs take the lead

Long QT syndrome (LQTS) is a detrimental arrhythmia syndrome mainly caused by dysregulated expression or aberrant function of ion channels. The major clinical symptoms of ventricular arrhythmia, palpitations and syncope vary among LQTS subtypes. Susceptibility to malignant arrhythmia is a result of delayed repolarisation of the cardiomyocyte action potential (AP). There are 17 distinct subtypes of LQTS linked to 15 autosomal dominant genes with monogenic mutations. However, due to the presence of modifier genes, the identical mutation may result in completely different clinical manifestations in different carriers. In this review, we describe the roles of various ion channels in orchestrating APs and discuss molecular aetiologies of various types of LQTS. We highlight the usage of patient-specific induced pluripotent stem cell (iPSC) models in characterising fundamental mechanisms associated with LQTS. To mitigate the outcomes of LQTS, treatment strategies are initially focused on small molecules targeting ion channel activities. Next-generation treatments will reap the benefits from development of LQTS patient-specific iPSC platform, which is bolstered by the state-of-the-art technologies including whole-genome sequencing, CRISPR genome editing and machine learning. Deep phenotyping and high-throughput drug testing using LQTS patient-specific cardiomyocytes herald the upcoming precision medicine in LQTS.

Cardiac Repolarization in Health and Disease

Abnormal cardiac repolarization is at the basis of life-threatening arrhythmias in various congenital and acquired cardiac diseases. Dysfunction of ion channels involved in repolarization at the cellular level are often the underlying cause of the repolarization abnormality. The expression pattern of the gene encoding the affected ion channel dictates its impact on the shape of the T-wave and duration of the QT interval, thereby setting the stage for both the occurrence of the trigger and the substrate for maintenance of the arrhythmia. Here we discuss how research into the genetic and electrophysiological basis of repolarization has provided us with insights into cardiac repolarization in health and disease and how this in turn may provide the basis for future improved patient-specific management.

The diagnostic role of T wave morphology biomarkers in congenital and acquired long QT syndrome: A systematic review

INTRODUCTION

QTc prolongation is key in diagnosing long QT syndrome (LQTS), however 25%-50% with congenital LQTS (cLQTS) demonstrate a normal resting QTc. T wave morphology (TWM) can distinguish cLQTS subtypes but its role in acquired LQTS (aLQTS) is unclear.

METHODS

Electronic databases were searched using the terms "LQTS," "long QT syndrome," "QTc prolongation," "prolonged QT," and "T wave," "T wave morphology," "T wave pattern," "T wave biomarkers." Whole text articles assessing TWM, independent of QTc, were included.

RESULTS

Seventeen studies met criteria. TWM measurements included T-wave amplitude, duration, magnitude, Tpeak-Tend, QTpeak, left and right slope, center of gravity (COG), sigmoidal and polynomial classifiers, repolarizing integral, morphology combination score (MCS) and principal component analysis (PCA); and vectorcardiographic biomarkers. cLQTS were distinguished from controls by sigmoidal and polynomial classifiers, MCS, QTpeak, Tpeak-Tend, left slope; and COG x axis. MCS detected aLQTS more significantly than QTc. Flatness, asymmetry and notching, J-Tpeak; and Tpeak-Tend correlated with QTc in aLQTS. Multichannel block in aLQTS was identified by early repolarization (ERD_30% ) and late repolarization (LRD_30% ), with ERD reflecting hERG-specific blockade. Cardiac events were predicted in cLQTS by T wave flatness, notching, and inversion in leads II and V₅ , left slope in lead V₆ ; and COG last 25% in lead I. T wave right slope in lead I and T-roundness achieved this in aLQTS.

CONCLUSION

Numerous TWM biomarkers which supplement QTc assessment were identified. Their diagnostic capabilities include differentiation of genotypes, identification of concealed LQTS, differentiating aLQTS from cLQTS; and determining multichannel versus hERG channel blockade.

A deep learning approach identifies new ECG features in congenital long QT syndrome

BACKGROUND

Congenital long QT syndrome (LQTS) is a rare heart disease caused by various underlying mutations. Most general cardiologists do not routinely see patients with congenital LQTS and may not always recognize the accompanying ECG features. In addition, a proportion of disease carriers do not display obvious abnormalities on their ECG. Combined, this can cause underdiagnosing of this potentially life-threatening disease.

METHODS

This study presents 1D convolutional neural network models trained to identify genotype positive LQTS patients from electrocardiogram as input. The deep learning (DL) models were trained with a large 10-s 12-lead ECGs dataset provided by Amsterdam UMC and externally validated with a dataset provided by University Hospital Leuven. The Amsterdam dataset included ECGs from 10000 controls, 172 LQTS1, 214 LQTS2, and 72 LQTS3 patients. The Leuven dataset included ECGs from 2200 controls, 32 LQTS1, and 80 LQTS2 patients. The performance of the DL models was compared with conventional QTc measurement and with that of an international expert in congenital LQTS (A.A.M.W). Lastly, an explainable artificial intelligence (AI) technique was used to better understand the prediction models.

RESULTS

Overall, the best performing DL models, across 5-fold cross-validation, achieved on average a sensitivity of 84 ± 2%, 90 ± 2% and 87 ± 6%, specificity of 96 ± 2%, 95 ± 1%, and 92 ± 4%, and AUC of 0.90 ± 0.01, 0.92 ± 0.02, and 0.89 ± 0.03, for LQTS 1, 2, and 3 respectively. The DL models were also shown to perform better than conventional QTc measurements in detecting LQTS patients. Furthermore, the performances held up when the DL models were validated on a novel external cohort and outperformed the expert cardiologist in terms of specificity, while in terms of sensitivity, the DL models and the expert cardiologist in LQTS performed the same. Finally, the explainable AI technique identified the onset of the QRS complex as the most informative region to classify LQTS from non-LQTS patients, a feature previously not associated with this disease.

CONCLUSIONS

This study suggests that DL models can potentially be used to aid cardiologists in diagnosing LQTS. Furthermore, explainable DL models can be used to possibly identify new features for LQTS on the ECG, thus increasing our understanding of this syndrome.

Congenital Long QT Syndrome

Congenital long QT syndrome (LQTS) encompasses a group of heritable conditions that are associated with cardiac repolarization dysfunction. Since its initial description in 1957, our understanding of LQTS has increased dramatically. The prevalence of LQTS is estimated to be ∼1:2,000, with a slight female predominance. The diagnosis of LQTS is based on clinical, electrocardiogram, and genetic factors. Risk stratification of patients with LQTS aims to identify those who are at increased risk of cardiac arrest or sudden cardiac death. Factors including age, sex, QTc interval, and genetic background all contribute to current risk stratification paradigms. The management of LQTS involves conservative measures such as the avoidance of QT-prolonging drugs, pharmacologic measures with nonselective β-blockers, and interventional approaches such as device therapy or left cardiac sympathetic denervation. In general, most forms of exercise are considered safe in adequately treated patients, and implantable cardioverter-defibrillator therapy is reserved for those at the highest risk. This review summarizes our current understanding of LQTS and provides clinicians with a practical approach to diagnosis and management.

Modeling genetic cardiac channelopathies using induced pluripotent stem cells - Status quo from an electrophysiological perspective

Long QT syndrome (LQTS), Brugada syndrome (BrS), and catecholaminergic polymorphic ventricular tachycardia (CPVT) are genetic diseases of the heart caused by mutations in specific cardiac ion channels and are characterized by paroxysmal arrhythmias, which can deteriorate into ventricular fibrillation. In LQTS3 and BrS different mutations in the SCN5A gene lead to a gain-or a loss-of-function of the voltage-gated sodium channel Nav1.5, respectively. Although sharing the same gene mutation, these syndromes are characterized by different clinical manifestations and functional perturbations and in some cases even present an overlapping clinical phenotype. Several studies have shown that Na⁺ current abnormalities in LQTS3 and BrS can also cause Ca²⁺-signaling aberrancies in cardiomyocytes (CMs). Abnormal Ca²⁺ homeostasis is also the main feature of CPVT which is mostly caused by heterozygous mutations in the RyR2 gene. Large numbers of disease-causing mutations were identified in RyR2 and SCN5A but it is not clear how different variants in the SCN5A gene produce different clinical syndromes and if in CPVT Ca²⁺ abnormalities and drug sensitivities vary depending on the mutation site in the RyR2. These questions can now be addressed by using patient-specific in vitro models of these diseases based on induced pluripotent stem cells (iPSCs). In this review, we summarize different insights gained from these models with a focus on electrophysiological perturbations caused by different ion channel mutations and discuss how will this knowledge help develop better stratification and more efficient personalized therapies for these patients.

Diagnosis, management and therapeutic strategies for congenital long QT syndrome

Congenital long QT syndrome (LQTS) is characterised by heart rate corrected QT interval prolongation and life-threatening arrhythmias, leading to syncope and sudden death. Variations in genes encoding for cardiac ion channels, accessory ion channel subunits or proteins modulating the function of the ion channel have been identified as disease-causing mutations in up to 75% of all LQTS cases. Based on the underlying genetic defect, LQTS has been subdivided into different subtypes. Growing insights into the genetic background and pathophysiology of LQTS has led to the identification of genotype–phenotype relationships for the most common genetic subtypes, the recognition of genetic and non-genetic modifiers of phenotype, optimisation of risk stratification algorithms and the discovery of gene-specific therapies in LQTS. Nevertheless, despite these great advancements in the LQTS field, large gaps in knowledge still exist. For example, up to 25% of LQTS cases still remain genotype elusive, which hampers proper identification of family members at risk, and it is still largely unknown what determines the large variability in disease severity, where even within one family an identical mutation causes malignant arrhythmias in some carriers, while in other carriers, the disease is clinically silent. In this review, we summarise the current evidence available on the diagnosis, clinical management and therapeutic strategies in LQTS. We also discuss new scientific developments and areas of research, which are expected to increase our understanding of the complex genetic architecture in genotype-negative patients, lead to improved risk stratification in asymptomatic mutation carriers and more targeted (gene-specific and even mutation-specific) therapies.

Electrocardiographic T-wave morphology and risk of mortality

BACKGROUND

Electrocardiographic T-wave morphology is used in drug safety studies as an adjunct to the QT_c interval, but few measurements of T-wave morphology can be interpreted in clinical practice. Morphology combination score (MCS) is a combination of T-wave flatness/peakedness, asymmetry, and notching, enabling easy visual assessment of T-wave morphology. We aimed to test the association between T-wave morphology, quantified by MCS, and mortality.

METHODS

We included electrocardiograms recorded in 2001-2011 from 342,294 primary care patients. Using Cox regression, we evaluated the association between MCS, cardiovascular death, and all-cause mortality, adjusting for heart rate, QT_c, QT-prolonging drugs, diabetes, ischemic heart disease, hypertension, and congestive heart failure.

RESULTS

270,039 individuals (44% men, median age 55 [inter-quartile range: 42-67 years]) were included and followed for a median of 9.3 years, during which time 13,489 (5.0%) died from cardiovascular causes and 50,481 (18.7%) from any cause. High values of MCS (i.e. asymmetric, flattened, and/or notched T waves) were associated with an adjusted mortality Hazard Ratio of 1.75 (95% CI 1.62-1.89) and 1.61 (1.43-1.92) for women and men, respectively. Low values of MCS (i.e. peaked and symmetric T waves) were associated with a Hazard Ratio of 1.18 (1.08-1.28) and 1.71 (1.48-1.98) for women and men, respectively.

CONCLUSIONS

In a large primary care population, we found that T-wave asymmetry, flatness, and notching provided prognostic information on mortality independent of heart rate, QTc, and baseline comorbidities.

Computational Diagnostic Techniques for Electrocardiogram Signal Analysis

Cardiovascular diseases (CVDs), including asymptomatic myocardial ischemia, angina, myocardial infarction, and ischemic heart failure, are the leading cause of death globally. Early detection and treatment of CVDs significantly contribute to the prevention or delay of cardiovascular death. Electrocardiogram (ECG) records the electrical impulses generated by heart muscles, which reflect regular or irregular beating activity. Computer-aided techniques provide fast and accurate tools to identify CVDs using a patient's ECG signal, which have achieved great success in recent years. Latest computational diagnostic techniques based on ECG signals for estimating CVDs conditions are summarized here. The procedure of ECG signals analysis is discussed in several subsections, including data preprocessing, feature engineering, classification, and application. In particular, the End-to-End models integrate feature extraction and classification into learning algorithms, which not only greatly simplifies the process of data analysis, but also shows excellent accuracy and robustness. Portable devices enable users to monitor their cardiovascular status at any time, bringing new scenarios as well as challenges to the application of ECG algorithms. Computational diagnostic techniques for ECG signal analysis show great potential for helping health care professionals, and their application in daily life benefits both patients and sub-healthy people.

Computer versus cardiologist: Is a machine learning algorithm able to outperform an expert in diagnosing a phospholamban p.Arg14del mutation on the electrocardiogram?

BACKGROUND

Phospholamban (PLN) p.Arg14del mutation carriers are known to develop dilated and/or arrhythmogenic cardiomyopathy, and typical electrocardiographic (ECG) features have been identified for diagnosis. Machine learning is a powerful tool used in ECG analysis and has shown to outperform cardiologists.

OBJECTIVES

We aimed to develop machine learning and deep learning models to diagnose PLN p.Arg14del cardiomyopathy using ECGs and evaluate their accuracy compared to an expert cardiologist.

METHODS

We included 155 adult PLN mutation carriers and 155 age- and sex-matched control subjects. Twenty-one PLN mutation carriers (13.4%) were classified as symptomatic (symptoms of heart failure or malignant ventricular arrhythmias). The data set was split into training and testing sets using 4-fold cross-validation. Multiple models were developed to discriminate between PLN mutation carriers and control subjects. For comparison, expert cardiologists classified the same data set. The best performing models were validated using an external PLN p.Arg14del mutation carrier data set from Murcia, Spain (n = 50). We applied occlusion maps to visualize the most contributing ECG regions.

RESULTS

In terms of specificity, expert cardiologists (0.99) outperformed all models (range 0.53-0.81). In terms of accuracy and sensitivity, experts (0.28 and 0.64) were outperformed by all models (sensitivity range 0.65-0.81). T-wave morphology was most important for classification of PLN p.Arg14del carriers. External validation showed comparable results, with the best model outperforming experts.

CONCLUSION

This study shows that machine learning can outperform experienced cardiologists in the diagnosis of PLN p.Arg14del cardiomyopathy and suggests that the shape of the T wave is of added importance to this diagnosis.