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.

OUP accepted manuscript

Aims

Genetic testing is recommended in specific inherited heart diseases but its role remains unclear and it is not currently recommended in unexplained cardiac arrest (UCA). We sought to assess the yield and clinical utility of genetic testing in UCA using whole-exome sequencing (WES).

Methods and results

Survivors of UCA requiring external defibrillation were included from the Cardiac Arrest Survivor with Preserved Ejection fraction Registry. Whole-exome sequencing was performed, followed by assessment of rare variants in previously reported cardiovascular disease genes. A total of 228 UCA survivors (mean age at arrest 39 ± 13 years) were included. The majority were males (66%) and of European ancestry (81%). Following advanced clinical testing at baseline, the likely aetiology of cardiac arrest was determined in 21/228 (9%) cases. Whole-exome sequencing identified a pathogenic or likely pathogenic (P/LP) variant in 23/228 (10%) of UCA survivors overall, increasing the proportion of ‘explained’ cases from 9% only following phenotyping to 18% when combining phenotyping with WES. Notably, 13 (57%) of the 23 P/LP variants identified were located in genes associated with cardiomyopathy, in the absence of a diagnosis of cardiomyopathy at the time of arrest.

Conclusions

Genetic testing identifies a disease-causing variant in 10% of apparent UCA survivors. The majority of disease-causing variants was located in cardiomyopathy-associated genes, highlighting the arrhythmogenic potential of such variants in the absence of an overt cardiomyopathy diagnosis. The present study supports the use of genetic testing including assessment of arrhythmia and cardiomyopathy genes in survivors of UCA.

Level of processing's effect on episodic retrieval following traumatic brain injury in the elderly: An event-related potential study

Individuals who have sustained a traumatic brain injury (TBI) can suffer from episodic memory impairments. Until now, the neural correlates underlying episodic retrieval in individuals with TBI remained scarce, particularly in older adults. We aimed to fill this gap by recording event-related potentials during an old/new episodic recognition task in 26 older adults, 13 healthy and 13 with TBI. The task manipulated the level of processing in encoding with the use of semantic organizational strategies (deep guided, deep self-guided, and shallow encoding). For all encoding conditions, behavioral data analyses on the discrimination rate indicated that older adults with TBI were globally impaired compared with healthy older adults. The electrophysiological results indicated that the left-parietal effect was larger in the deep guided condition than in the shallow condition. In addition, the results show that the mid-frontal and left-parietal positive old/new effects were absent in both groups. The main findings are the observation, in the control group only, of an early frontal old/new effect (P200; 150-300 ms) and of a late frontal old/new effect on the left hemisphere, only in the Spontaneous condition. Together, results suggest an impairment of the allocation of attentional resources and working memory necessary for retrieving and monitoring items in the elderly with TBI.

Missense variants in the spectrin repeat domain of DSP are associated with arrhythmogenic cardiomyopathy: A family report and systematic review

Rare loss of function variants in DSP, which codes for the desmosomal protein desmoplakin, have been implicated in dilated and arrhythmogenic right ventricular cardiomyopathies. We present a family with arrhythmogenic cardiomyopathy associated with a novel missense variant in DSP (NM_004415.4): c.877G>A, p.(Glu293Lys). The phenotype is characterized by predominant involvement of the left ventricle with systolic dysfunction, fibrosis, and life-threatening arrhythmias. We performed a systematic review of literature collecting all cardiomyopathy cases with rare missense variants in DSP. We demonstrate that the distribution of missense variants across the protein domains in cardiomyopathy cases differs from that in gnomAD (p = .04), with a case enrichment of rare missense variants in the spectrin repeat domain (36/78 [46%] in cases vs. 449/1495 [30%] in gnomAD; p = .004). Our findings highlight the predominance of cardiac arrhythmia and left ventricular involvement in desmoplakin cardiomyopathy and pinpoint to a potential mutation hotspot in DSP thereby facilitating missense variant interpretation in the diagnostic setting.

A tRNS investigation of the sensory representation of time

The understanding of the mechanisms underlying the representation of temporal intervals in the range of milliseconds/seconds remains a complex issue. Different brain areas have been identified as critical in temporal processing. The activation of specific areas is depending on temporal range involved in the tasks and on the modalities used for marking time. Here, for the first time, transcranial random noise stimulation (tRNS) was applied over the right posterior parietal (P4) and right frontal (F4) cortex to investigate their role in intra- and intermodal temporal processing involving brief temporal intervals (<1 sec). Eighty University students performed a time bisection task involving standard durations lasting 300 ms (short) and 900 ms (long). Each empty interval to be judged was marked by two successive brief visual (V) or auditory (A) signals defining four conditions: VV, VA, AV or AA. Participants were assigned to one of these four conditions. Half of the participants received tRNS over P4 and half over F4. No effect of stimulation was observed on temporal variability (Weber ratio). However, participants that were stimulated over P4 overestimated temporal intervals in the random condition compared to the sham condition. In addition to showing an effect of tRNS on perceived duration rather than on temporal variability, the results of the present study confirm that the right posterior parietal cortex is involved in the processing of time intervals and extend this finding to several sensory modality conditions.

Primary laparoscopic and open repair of paraesophageal hernias: a comparison of short-term outcomes

The choice of the optimal surgical approach for repairing paraesophaeal hernias (PEH) is debated. Our objective is to evaluate the short-term outcomes of primary laparoscopic and open repairs of PEH performed in the Calgary Health Region. A retrospective review of all patients undergoing repair of PEH between October 1999 and February 2005 was performed. The outcome measures evaluated included intra-operative parameters and post-operative variables, mortality rates, recurrence rates and patient satisfaction. A total of 93 patients underwent either a laparoscopic (n = 46) or open (n = 47) primary PEH repair. The laparoscopic approach was associated with a longer mean operative time (3.1 +/- 1.2 hours vs. 2.5 +/- 0.7 hours, P = 0.005) but resulted in a shorter overall hospital stay (5 days [2-16 days]vs. 10 days [5-24 days]; P < 0.001), and fewer post-operative complications (10/46 [22%]vs. 25/47 [53%]P = 0.002). Although the follow-up was short (laparoscopic 16 months; open 18 months), a 9% recurrence rate was reported with both approaches. Patient satisfaction using the Gastroesophageal Disease Health-Related Quality Of Life questionnaire was similar in both groups (P = 0.861) with most patients reporting excellent outcomes (laparoscopic: 32/36 [89%]; open 27/35 [77%]). Our review suggests that the laparoscopic approach is safe with shorter hospital stay and recovery. Although early follow-up suggests that recurrence rates and patient satisfaction are similar, long-term follow-up is required to determine whether the laparoscopic approach will become the procedure of choice.

[Perturbed consciousness in schizophrenia: an evaluation of C.D. Frith's model]

INTRODUCTION

While many neurocognitive models of schizophrenia coexist, a lot of attention has been centered on C.D. Frith's model over the past few years, especially in regard to its parsimony.

BACKGROUND

The aim of this paper is to examine its validity. The model relies on the assumption that there are two main components: self-monitoring and monitoring. The first permits one to keep consciousness of personal goals and intentions with metarepresentations. Losing consciousness of personal goals would be the source of schizophrenics' avolition and losing consciousness of personal intentions would generate reference ideas. The second component refers to the so-called "theory of mind", which is the monitoring of others' mental content (knowledge and intentions). Disturbing monitoring would cause schizophrenics persecution disillusions, third order persecutions or speech content disorders.

LITERATURE FINDINGS

After reviewing the empirical and theoretical bases of Frith's model, strengths and weaknesses are highlighted, in particular by contrasting Hardy-Baylé's and Abu-Akel's theoretical proposals. For explaining the monitoring impairments of schizophrenics, Hardy-Baylé's model emphasizes the executive functioning defect, while Abu-Akel's model proposes a "hyper theory of mind" where too many hypotheses would lead to misattributions. In addition, several criticisms of Frith's model are examined, particularly those voiced by phenomenologists who underline its reductionism presupposition and argue that the underlying cognitive conception of the "theory of mind" neglects the fundamental intersubjectivity issue. In addition, Gallagher points out that monitoring is a tautological concept and that intention is not like thinking inherent to behaviour.

CONCLUSION

Frith's model validity is finally discussed at large in the light of these criticisms and competing models, and it is concluded that the self-monitoring part of the model needs to be redefined and that the measurement of the "theory of mind" has to be standardized.

Discriminating time intervals presented in sequences marked by visual signals

This article presents the results of three experiments on the discrimination of time intervals presented in sequences marked by brief visual signals. In Experiment 1A (continuous condition), the participants had to indicate whether, in a series of 2-4 intervals marked by 3-5 visual signals, the last interval was shorter or longer than the previous one(s). In Experiment 1B (discontinuous condition), the participants indicated whether, in a presentation of two series of 1-3 intervals, with each series being marked by 2-4 signals, the intervals of the second sequence were shorter or longer than those of the first. Whenever one, two, or three standard intervals were presented, the difference threshold was as high at 150 msec as it was at 300 msec with the continuous method but increased monotonically from 150 to 900 msec with the discontinuous method. With both methods, the increase was well described by Weber's law--the Weber fraction was roughly constant--between 600 and 900 msec (Experiment 2), whereas between 900 and 1,200 msec (Experiment 3), the Weber fraction increased.

From physical time to the first and second moments of psychological time

After examination of the status of time in experimental psychology and a review of related major texts, 2 opposite approaches are presented in which time is either unified or fragmented. Unified time perception views, usually guided by Weber's law, are embodied in various models. After a brief review of old models and a description of the major contemporary models of time perception, views on fragmented time perception are presented as challenges for any unified time view. Fragmentation of psychological time emerges from (a) disruptions of the Weber function, which are caused by the types of interval presentation, by extensive practice, and by counting explicitly or not; and (b) modulations of time sensitivity and perceived duration by attention and interval structures. Weber's law is a useful guide for studying psychological time, but it is also reasonable to assume that more than one so-called central timekeeper could contribute to perceiving time.

Manual laterality and hitting performance in major league baseball

Asymmetrical hand function was examined in the context of expert sports performance: hitting in professional baseball. An archival study was conducted to examine the batting performance of all Major League Baseball players from 1871 to 1992, focusing on those who batted left (n = 1,059) to neutralize the game asymmetry. Among them, left-handers (n = 421) were more likely to hit with power and to strike out than right-handers (n = 638). One possible account, based on the idea of hand dominance and an analogy to tennis, is that batting left involves a double-handed forehand for left-handers and a weaker and more reliable double-handed backhand for right-handers. The results are also interpretable in the light of Y. Guiard's (1987) kinematic chain model of a between-hands asymmetrical division of labor, which provides a detailed account of why left batting is optimal for left-handers.

Manual laterality and hitting performance in major league baseball

Asymmetrical hand function was examined in the context of expert sports performance: hitting in professional baseball. An archival study was conducted to examine the batting performance of all Major League Baseball players from 1871 to 1992, focusing on those who batted left (n = 1,059) to neutralize the game asymmetry. Among them, left-handers (n = 421) were more likely to hit with power and to strike out than right-handers (n = 638). One possible account, based on the idea of hand dominance and an analogy to tennis, is that batting left involves a double-handed forehand for left-handers and a weaker and more reliable double-handed backhand for right-handers. The results are also interpretable in the light of Y. Guiard's (1987) kinematic chain model of a between-hands asymmetrical division of labor, which provides a detailed account of why left batting is optimal for left-handers.

Sensory effects on judgments of short time-intervals

Two experiments were conducted to test the effect of nontemporal factors on duration discrimination. In Exp. 1, a forced-choice adaptive procedure with a standard duration of 400 or 800 ms was employed. It was shown that, for both auditory and visual modes, the discrimination is better with empty intervals (a silent period between two brief signals) than with filled intervals (a continuous signal), but only with shorter durations. In a second experiment, where intervals of the same duration range were employed but were presented with a single-stimulus method, discrimination was better with empty than with filled intervals, and this effect applied to both ranges of duration and both sensory modes. In both experiments, discrimination was better in the auditory than in the visual mode. These data are discussed in the context of current models of timing mechanisms.

The role of endoscopy in the staging and management of lung metastases

Endoscopic evaluation of the patient with lung metastases takes on many forms depending upon the extent of disease and the intent of treatment, be that curative or palliative. Thorascopy, and occasionally mediastinoscopy, may be helpful in assessing operability in patients with extensive disease on a preoperative computed tomography scan. However, when in doubt, exploration is always indicated in the young, good risk patient. Palliative efforts usually concern airway obstruction and malignant effects. A variety of technologies, including laser, brachytherapy, and endoluminal stents, helps manage symptomatic bronchial or tracheal lesions.

Judgments of the duration of visually marked empty time intervals: linking perceived duration and sensitivity

The capability of subjects to categorize (as short or long) visually marked empty time intervals was investigated in three experiments. Two visual signals, located 18 degrees to the left (L) and to the right (R) of a fixation point in the visual field, established four marking conditions, two unilaterally presented (L-L and R-R) and two bilaterally presented (L-R and R-L). In Experiments 1 and 2, the results show that discrimination is better with unilateral sequences than with bilateral sequences and that the perceived duration is longer with an L-R than with an R-L sequence. In addition, Experiment 2 shows that, in comparison with a condition in which Markers 1 and 2 remain identical for a complete session, varying the markers from trial to trial does not decrease discrimination. Also, Experiment 2 shows that discrimination is better when both visual markers are presented at fovea than it is in the unilateral conditions. Experiment 3 shows that bilateral intervals are perceived as being longer and are better discriminated than are intervals marked by an intermodal sequence (auditory-visual or visual-auditory). The general discussion reports the implications of having different perceived duration and sensitivity levels, in various marker-type conditions, for an internal-clock hypothesis. Some implications of these results for a lateralized-timer hypothesis are also discussed.

Markers' influence on the duration discrimination of intermodal intervals

The effects of sensory signal characteristics on the duration discrimination of intermodal intervals was investigated in three experiments. Temporal intervals were marked by either the successive presentation of a visual then auditory signal (VA), or by the successive presentation of an auditory then visual signal (AV). The results indicated that (1) VA intervals are generally easier to discriminate than are AV intervals, but this effect depends on the range of duration studied; (2) AV intervals are perceived as longer than VA intervals for durations ranging from 250 to 750 msec; (3) the intensity of the visual markers for both AV and VA intervals does not affect the discrimination; and (4) the perceived duration of an intermodal interval is influenced by the length of the first and second markers. The results are mainly interpreted in terms of (1) a sensory trace left by visual and auditory signals and (2) the detection of these signals.

Controlled attention sharing influences time estimation

A seminal attentional model of time estimation predicts that subjective duration will be positively correlated to the amount of attention given to temporal processing. This prediction holds under prospective conditions, in which one is forewarned that judgments of time will be asked, in contrast to retrospective conditions, in which such judgments are required after the relevant period without any prior warning. In three experiments, an attention-sharing method was used. Subjects were asked to control the amount of attention that they devoted to one or the other component of a dual-task paradigm. The first experiment involved word categorization and reproduction of duration. The following experiments, based on signal detection theory, required discrimination of both the duration and the intensity of a single stimulus, in the visual (Experiment 2) or the auditory (Experiment 3) modality. The results indicate that when the attention is directly controlled by the subject, the subjective duration shortens as the amount of attention devoted to the temporal task diminishes. The implications of these results for the possible existence of an internal timer are considered.

Duration discrimination of empty and filled intervals marked by auditory and visual signals

Experiments 1 and 2 compared, with a single-stimulus procedure, the discrimination of filled and empty intervals in both auditory and visual modalities. In Experiment 1, in which intervals were about 250 msec, the discrimination was superior with empty intervals in both modalities. In Experiment 2, with intervals lasting about 50 msec, empty intervals showed superior performance with visual signals only. In Experiment 3, for the auditory modality at 250 msec, the discrimination was easier with empty intervals than with filled intervals with both the forced-choice (FC) and the single stimulus (SS) modes of presentation, and the discrimination was easier with the FC than with the SS method. Experiment 4, however, showed that at 50 and 250 msec, with a FC-adaptive procedure, there were no differences between filled and empty intervals in the auditory mode; the differences observed with the visual mode in Experiments 1 and 2 remained significant. Finally, Experiment 5 compared differential thresholds for four marker-type conditions, filled and empty intervals in the auditory and visual modes, for durations ranging from .125 to 4 sec. The results showed (1) that the differential threshold differences among marker types are important for short durations but decrease with longer durations, and (2) that a generalized Weber's law generally holds for these conditions. The results as a whole are discussed in terms of timing mechanisms.

Production of time intervals from segmented and nonsegmented inputs

One important factor influencing the accuracy of a timing estimate is the counting activity that a human subject may adopt. In the present study, the usefulness of this activity is evaluated with a strategy whereby subjects are presented segmented and nonsegmented intervals, before they start to produce a series of these intervals, using a finger-tapping procedure. The results are mainly analyzed in the light of Killeen and Weiss's (1987) model, which addressed this question of counting. The results revealed that (1) a scalar property gives a better description of the pacemaker activity than does a Poisson process, and (2) an optimal timing performance would be reached with the utilization of subintervals with an approximate value of 400 msec. Finally, the discussion also incorporates an analysis of the variability related to the motor component in a tapping task.

Judging the relative duration of multimodal short empty time intervals

Three experiments address the cause of the different performance levels found in time discrimination of empty intervals with durations near 250 msec. Performance differed according to the kind of sensory modality that marked the intervals. With a procedure in which the type of marker was randomized from trial to trial, it was shown that variability of discrimination judgments could not be attributed entirely to the variability of the criterion on which a judgment was based. Such a randomization slightly affects discrimination but provokes a reorganization related to marker conditions of the probabilities of judging an interval to be short or long. Moreover, it was shown that within intramodal conditions, physical characteristics of markers influence the discrimination performances. To account for the results generated with different marker-type intervals at 250 msec, we propose that two types of processor may be involved in duration discrimination: one is specifically related to a given sensory modality, whereas the other is aspecific and responsible for discrimination of intermodal intervals.