The Telemetric and Holter ECG Warehouse (THEW): the first three years of development and research

Heart failure classification using deep learning to extract spatiotemporal features from ECG

BACKGROUND

Heart failure is a syndrome with complex clinical manifestations. Due to increasing population aging, heart failure has become a major medical problem worldwide. In this study, we used the MIMIC-III public database to extract the temporal and spatial characteristics of electrocardiogram (ECG) signals from patients with heart failure.

METHODS

We developed a NYHA functional classification model for heart failure based on a deep learning method. We introduced an integrating attention mechanism based on the CNN-LSTM-SE model, segmenting the ECG signal into 2 to 20 s long segments. Ablation experiments showed that the 12 s ECG signal segments could be used with the proposed deep learning model for superior classification of heart failure.

RESULTS

The accuracy, positive predictive value, sensitivity, and specificity of the NYHA functional classification method were 99.09, 98.9855, 99.033, and 99.649%, respectively.

CONCLUSIONS

The comprehensive performance of this model exceeds similar methods and can be used to assist in clinical medical diagnoses.

Individualized QT interval (QTi) is a powerful diagnostic tool in long QT syndrome: results from a large validation study

Aims

Diagnosis of Long QT syndrome (LQTS) is based on prolongation of the QT interval corrected for heart rate (QTc) on surface ECG and genotyping. However, up to 25% of genotype positive patients have a normal QTc interval. We recently showed that individualized QT interval (QTi) derived from 24 h holter data and defined as the QT value at the intersection of an RR interval of 1,000 ms with the linear regression line fitted through QT-RR data points of each individual patient was superior over QTc to predict mutation status in LQTS families. This study aimed to confirm the diagnostic value of QTi, fine-tune its cut-off value and evaluate intra-individual variability in patients with LQTS.

Methods

From the Telemetric and Holter ECG Warehouse, 201 recordings from control individuals and 393 recordings from 254 LQTS patients were analysed. Cut-off values were obtained from ROC curves and validated against an in house LQTS and control cohort.

Results

ROC curves indicated very good discrimination between controls and LQTS patients with QTi, both in females (AUC 0.96) and males (AUC 0.97). Using a gender dependent cut-off of 445 ms in females and 430 ms in males, a sensitivity of 88% and specificity of 96% were achieved, which was confirmed in the validation cohort. No significant intra-individual variability in QTi was observed in 76 LQTS patients for whom at least two holter recordings were available (483 ± 36 ms vs. 489 ± 42 ms, p = 0.11).

Conclusions

This study confirms our initial findings and supports the use of QTi in the evaluation of LQTS families. Using the novel gender dependent cut-off values, a high diagnostic accuracy was achieved.

Accurate QT correction method from transfer entropy

Background

The QT interval in the electrocardiogram (ECG) is a fundamental risk measure for arrhythmic adverse cardiac events. However, the QT interval depends on the heart rate and must be corrected accordingly. The present QT correction (QTc) methods are either simple models leading to under- or overcorrection, or impractical in requiring long-term empirical data. In general, there is no consensus on the best QTc method.

Objective

We introduce a model-free QTc method-AccuQT-that computes QTc by minimizing the information transfer from R-R to QT intervals. The objective is to establish and validate a QTc method that provides superior stability and reliability without models or empirical data.

Methods

We tested AccuQT against the most commonly used QT correction methods by using long-term ECG recordings of more than 200 healthy subjects from PhysioNet and THEW databases.

Results

AccuQT overperforms the previously reported correction methods: the proportion of false-positives is reduced from 16% (Bazett) to 3% (AccuQT) for the PhysioNet data. In particular, the QTc variance is significantly reduced and thus the RR-QT stability is increased.

Conclusion

AccuQT has significant potential to become the QTc method of choice in clinical studies and drug development. The method can be implemented in any device recording R-R and QT intervals.

Readily available ECG databases

Medical databases are an essential tool in health research that have become more frequently used in the last few decades alongside the growth and worldwide availability of the internet. Heart diseases are one of the most common health problems in the world with a vast amount of work being done on disease prevention and surveillance. With respect to arrhythmia detection, the electrocardiogram (ECG) is one of the most studied monitoring tools to date, generating large amounts of real-time data and giving rise to readily available ECG databases. In this research, we provide an overview of the most cited ECG databases with public/read access and discuss their characteristics with emphasis on the ECG records, as well as their use and applications carried out by other researchers around the world.

High-frequency autonomic modulation: a new model for analysis of autonomic cardiac control

Background and Purpose

Increase in high‐frequency beat‐to‐beat heart rate oscillations by torsadogenic hERG blockers appears to be associated with signs of parasympathetic and sympathetic co‐activation which cannot be assessed directly using classic methods of heart rate variability analysis. The present work aimed to find a translational model that would allow this particular state of the autonomic control of heart rate to be assessed.

Experimental Approach

High‐frequency heart rate and heart period oscillations were analysed within discrete 10 s intervals in a cohort of 200 healthy human subjects. Results were compared to data collected in non‐human primates and beagle dogs during pharmacological challenges and torsadogenic hERG blockers exposure, in 127 genotyped LQT1 patients on/off β‐blocker treatment and in subgroups of smoking and non‐smoking subjects.

Key Results

Three states of autonomic modulation, S1 (parasympathetic predominance) to S3 (reciprocal parasympathetic withdrawal/sympathetic activation), were differentiated to build a new model of heart rate variability referred to as high‐frequency autonomic modulation. The S2 state corresponded to a specific state during which both parasympathetic and sympathetic systems were coexisting or co‐activated. S2 oscillations were proportionally increased by torsadogenic hERG‐blocking drugs, whereas smoking caused an increase in S3 oscillations.

Conclusions and Implications

The combined analysis of the magnitude of high‐frequency heart rate and high‐frequency heart period oscillations allows a refined assessment of heart rate autonomic modulation applicable to long‐term ECG recordings and offers new approaches to assessment of the risk of sudden death both in terms of underlying mechanisms and sensitivity.

A novel technique to investigate the effect of ageing on ventricular repolarization characteristics in healthy and LQTS subjects

Ventricular repolarization(VR) characteristics is affected by ageing alongside several other factors like Heart rate(HR),respiration, modulation of autonomic nervous system, different drug effects, genetical factors affecting the cardiac ion channel characteristics, gender etc. Therefore, total VR variability (i.e. QT interval variability in surface ECG) consists of two components: one dependent on HR variability (HRV) and another independent of HRV. Analysis of QT interval variability (QTV) is crucial for both healthy and pathological conditions as increase in VR variability measured by QTV increases cardiac repolarization instability, which might lead to arrhythmogenesis. Analyzing the effect of ageing using a widely used measure of QTV (i.e. QTVI) is reported inconsistently in Healthy subjects whereas the same for Long QT Syndrome (LQTS) subjects is not widely reported. In this study, we propose a novel time domain measure from beat-tobeat QT-RR distribution to analyze how ageing affects VR in both Healthy and a group of genotyped LQTS1 subjects. A total of 139 Healthy subjects and 134 LQTS1 subjects of three different age groups (i.e. Young: age 20-35, Middle-aged: 40-55 and Old: age<;60) were analyzed for this study. The proposed measure is also compared with other existing widely used measures of QTV like SDQT and QTVI in differentiating different age groups. The proposed measure stands out to be more discriminatory than other existing variability measures of QT interval.

An evaluation of multiple algorithms for the measurement of the heart rate corrected JTpeak interval

Interest in the effects of drugs on the heart rate-corrected JTpeak (JTpc) interval from the body-surface ECG has spawned an increasing number of scientific investigations in the field of regulatory sciences, and more specifically in the context of the Comprehensive in vitro Proarrhythmia Assay (CiPA) initiative. We conducted a novel initiative to evaluate the role of automatic JTpc measurement technologies by comparing their ability to distinguish multi- from single-channel blocking drugs. A set of 5232 ECGs was shared by the FDA (through the Telemetric and Holter ECG Warehouse) with 3 ECG device companies (AMPS, Mortara, and Philips). We evaluated the differences in drug-concentration effects on these measurements between the commercial and the FDA technologies. We provide a description of the drug-induced placebo-corrected changes from baseline for dofetilide, quinidine, ranolazine, and verapamil, and discuss the various differences across all technologies. The results revealed only small differences between measurement technologies evaluated in this study. It also confirms that, in this dataset, the JTpc interval distinguishes between multi- and single-channel (hERG) blocking drugs when evaluating the effects of dofetilide, quinidine, ranolazine, and verapamil. However, in the case of quinidine and dofetilide, we noticed a poor consistency across technologies because of the lack of standard definitions for the location of the peak of the T-wave (T-apex) when the T-wave morphology is abnormal.

"QT clock" to improve detection of QT prolongation in long QT syndrome patients

BACKGROUND

The QT interval is a risk marker for cardiac events such as torsades de pointes. However, QT measurements obtained from a 12-lead ECG during clinic hours may not capture the full extent of a patient's daily QT range.

OBJECTIVE

The purpose of this study was to evaluate the utility of 24-hour Holter ECG recording in patients with long QT syndrome (LQTS) to identify dynamic changes in the heart rate-corrected QT interval and to investigate methods of visualizing the resulting datasets.

METHODS

Beat-to-beat QTc (Bazett) intervals were automatically measured across 24-hour Holter recordings from 202 LQTS type 1, 89 type 2, and 14 type 3 genotyped patients and a reference group of 200 healthy individuals. We measured the percentage of beats with QTc greater than the gender-specific threshold (QTc ≥470 ms in women and QTc ≥450 ms in men). The percentage of beats with QTc prolongation was determined across the 24-hour recordings.

RESULTS

Based on the median percentage of heartbeats per patient with QTc prolongation, LQTS type 1 patients showed more frequent QTc prolongation during the day (~3 PM) than they did at night (~3 AM): 97% vs 48%, P ~10(-4) for men, and 68% vs 30%, P ~10(-5) for women. LQTS type 2 patients showed less frequent QTc prolongation during the day compared to nighttime: 87% vs 100%, P ~10(-4) for men, and 62% vs 100%, P ~10(-3) for women.

CONCLUSION

In patients with genotype-positive LQTS, significant differences exist in the degree of daytime and nocturnal QTc prolongation. Holter monitoring using the "QT clock" concept may provide an easy, fast, and accurate method for assessing the true personalized burden of QTc prolongation.