Impaired heart rate variability one and six months post acute COVID-19

Background

Long COVID-19 syndrome is an increasingly recognized problem. Post-infectious cardiac autonomic dysfunction is commonly reported. This study aims to evaluate autonomic dysfunction by means of Heart rate variability (HRV) on post-COVID-19 patients.

Methods

Hospitalized patients for COVID-19 (either at the medical ward or Intensive Care Unit (ICU)) were followed up at 1 and 6 months after hospital discharge. Medical history and clinical information were collected. HRV was assessed by 24-hour ambulatory electrocardiography Holter, with the measure of the standard deviation of normal RR intervals in 24 h, ms (SDNN). The comparison was conducted with age and sex-matched non-COVID-19 controls.

Results

Thirty-four patients hospitalized with COVID-19 (20.6% admitted in ICU) were examined 1-month and 6-months post-hospital discharge. SDNN was significantly (p<0.001) reduced in the COVID-19 group (111±23 ms) compared to the control subjects (152±24 ms) 1-month after discharge. Subgroup analysis between COVID-19 group revealed that ICU subjects presented significantly (p<0.001) reduced SDNN compared to the medical ward, respectively (83±20 ms vs. 118±17 ms). At 6-months, an improvement was noted at SDNN 24h (6-month: 133±24 vs. control: 151±24 ms, p=0.004; 1-month: 111±23 ms vs. 6-month: 133±24 ms, p<0.001). Also at 6-months, ICU subjects noted significantly (p=0.003) reduced SDNN 24h compared to medical ward subjects (107±17 ms vs. 140±20 ms). On the 6-months follow-up, 32% of the subjects had “long-COVID-19” symptoms. Subjects with long COVID-19 symptoms had low SDNN values (“long-COVID-19”: 112±17 ms vs. non-“long-COVID-19”: 142±20 ms, p=0.001)

Conclusion

Patients hospitalized for COVID-19 have reduced SDNN, at one month post-hospital discharge which is improved at the six months follow-up. These findings emphasize the increased sympathetic drive activity in the post-acute COVID-19 phase and imply a link between autonomic dysfunction and long COVID-19.

Funding Acknowledgement

Type of funding sources: None.

Ventricular-arterial coupling impairment in patients recovered from COVID-19

Background

Myocardial and vascular alterations among post-COVID-19 patients are observed. The coupling between arterial stiffness with left ventricular (LV) myocardial function (ventricular-arterial coupling, VAC) is an important determinant of cardiovascular performance and cardiac energetics. The aim of the study was to investigate the spectrum of cardiac and vascular abnormalities at mid-term follow-up in post-COVID-19 patients.

Methods

We enrolled 25 hospitalized patients for COVID-19, at one and six months after hospital discharge. The ratio (PWV/GLS) of carotid-femoral pulse wave velocity (cf-PWV), as a marker of arterial stiffness, to global longitudinal strain (LV-GLS), as a marker of left ventricular performance, was estimated as a marker of arterial elastance/left ventricular elastance index the long-term. The comparison was conducted with age and sex-matched non-COVID-19 controls.

Results

There was no difference in age (56.8±11.6 y vs. 57.4±9.5 y; p=0.85) and male sex (64% vs. 68%; P=0.76) between post-COVID-19 and control subjects respectively. At one-month follow-up, significant impairment was noted between post-COVID-19 and control subjects regarding: VAC (−0.71±0.24 m/s% vs. −0.44±0.11 m/sec%; p&lt;0.001), LV-GLS (−17.9±3.1% vs. −21.9±2.7%; p&lt;0.001), cf-PWV (12.3±3.5 m/s vs. 9.6±1.9; p&lt;0.001). At six-month follow-up, an improvement was observed but there still was significant difference between post-COVID-19 and control subjects in: VAC (−0.62±0.19 m/sec% vs. −0.44±0.11 m/sec%; p&lt;0.001), LV-GLS (−19.3±2.9% vs. −21.9±2.7%; p=0.001), cf-PWV (11.7±2.7 m/s vs. 9.6±1.9 m/s; p=0.001). Moreover, it was observed at 1-month: VAC adverse correlation with the levels of IL-6 (r=−0.54; p&lt;0.001), CRP (−0.71; p=0.011) and at 6-months: IL-6 (r=−0.47; p=0.003), CRP (−0.56; p=0.007).

Conclusion

Ventricular-arterial coupling is impaired 6 months following COVID-19 highlighting the possible effects of SARS-CoV-2 infection in left ventricular mechanics and performance.

Funding Acknowledgement

Type of funding sources: None.

The role of cardiometabolic risk factors and endothelial dysfunction in serum albumin levels and capillary leak syndrome of patients with COVID-19

Background

Growing evidence focuses on the role of hypoalbuminemia in the COVID-19 course and the role of vascular inflammation in the progression to Capillary Leak Syndrome (CLS). CLS may be mediated by a derangement of endothelial barrier following vascular endothelial dysfunction. We investigated the role of cardiometabolic risk factors in the association of hypoalbuminemia with endothelial dysfunction of hospitalized COVID-19 patients.

Methods

In this cross-sectional study, patients hospitalized for COVID-19 at the medical ward or Intensive Care Unit (ICU) were enrolled. Medical history and laboratory examinations were collected while the endothelial function was assessed by brachial artery flow-mediated dilation (FMD) between the first 24–72 hours of their admission to the hospital. According to the body mass index, history of hypertension, dyslipidemia, and diabetes mellitus, COVID-19 patients were categorized in those with Cardiometabolic Risk Factors (CRFact) or without CRFact (no-CRFact). From the study population, we excluded subjects with established cardiovascular disease.

Results

Sixty-six patients with COVID-19 (37% admitted in ICU) were recruited. From the study population, 41 were in the group of CRFact and 25 in the no-CRFact. Patients with CFRact were older (65±9 years vs. 53±14 years, p<0.001), had more impaired FMD (1.16±2.13% vs. 2.60±2.44%, p=0.01), and lower serum albumin levels (3.10±0.68 g/dL vs. 3.52±0.26 g/dL, p=0.006) compared to the no-CRFact group. Between CRFact and no-CRFact, there was no difference in CRP and IL-6 levels. Interestingly, serum albumin in patients with CRFact was significantly lower than the lower reference limit (LRL) (=3.5 g/dl) of albumin (p=0.001), while no such finding was noted in subjects with no CRFact (p=0.64). Furthermore, regression analysis revealed that, even after adjustment for age, the presence of CRFact was associated with decreased serum albumin levels by 0.31mg/dl (95% CI 0.08 to 0.63, p=0.04). In the CRFact population, there was a correlation of albumin with FMD (R=0.29, p=0.05) and an inverse correlation with CRP (rho=−0.48, p=0.02) and IL-6 (rho=−0.66, p<0.001), while in the no-CRFact group no such correlation were observed (p=NS for all).

Conclusion

COVID-19 patients with cardiometabolic risk factors present with low serum albumin levels early at the course of the disease, which may be driven by endothelial dysfunction and vascular inflammation. This data gives insights into the potential association of a dysfunctional endothelial layer and the progression to capillary leak syndrome.

Funding Acknowledgement

Type of funding sources: None.

Inside the “brain” of an artificial neural network: an interpretable deep learning approach to paroxysmal atrial fibrillation diagnosis from electrocardiogram signals during sinus rhythm

Background

With the ongoing, rapid advances in Deep Learning (DL), such solutions can now detect medical conditions even invisible to the human eye. In this direction, efforts have been made to develop DL algorithms that diagnose paroxysmal atrial fibrillation (PAF) from electrocardiogram (ECG) signals in sinus rhythm (SR). However, many of the available approaches function as “black boxes”, with physicians unable to understand and trust their predictions.

Purpose

To train a DL model to detect PAF patients while in SR and apply an algorithm that interprets and visualises its decisions.

Methods

We obtained ECG samples from PAF and non-PAF patients during SR, from the PAF Prediction Challenge Database. After discarding unannotated samples and augmenting the sample size (by dividing each signal into 30-second segments), we split the whole dataset into a train (68%), a validation (16%) and a test (16%) set. No pair of samples belonging to different sets originated from the same patient. We trained the InceptionTime neural network on the train/validation sets and tested on the “unseen” test set after “hiding” the correct answers. Its performance was evaluated with the following metrics: Accuracy, f1-score, precision and recall (sensitivity). After repeating this process 20 times, we obtained a distribution for each score. Finally, we adjusted the Grad-CAM interpretation algorithm to our data and used it to visualise the areas perceived as important by the model.

Results

After pre-processing, 4,080, 30-second, two-lead ECG signals were allocated to the train set, 960 to the validation and 960 to the test set. Each subset contained an equal number of PAF and non-PAF samples. After repeated training and testing, we obtained a median accuracy of 0.84 (interquartile range, IQR: 0.66–0.88), an f1-score of 0.82 (IQR: 0.68–0.88) and a median precision and recall equal to 0.93 (IQR: 0.67–0.99) and 0.77 (IQR: 0.68–0.93), respectively. The Grad-CAM technique highlighted the ECG areas of interest that led to each decision. We selected and present both PAF-positive and -negative samples, perceived either correctly or falsely. Interestingly, correct model decisions tend to focus on the P-wave, while false ones fixate on other regions.

Conclusions

Although a pilot study with considerable limitations (small sample size, disregard of possible confounding due to comorbidities or other factors), this work shows how DL can be employed to distinguish between PAF and non-PAF patients from SR ECG samples, and confirms the potential of DL-enabled approaches to offer novel diagnostic capabilities. Most importantly, our effort provides a comprehensible, visual interpretation of the model's decisions. Demystifying DL behaviour can, not only improve such efforts by explaining false decisions, but also cultivate trust among clinicians and, possibly, point out directions for future research, since we can now see through the magnifying lens of a neural network.

Funding Acknowledgement

Type of funding sources: None.

Optimising and validating deep learning approaches for diagnosing atrial fibrillation from few-lead ambulatory electrocardiogram signals

Funding Acknowledgements

Type of funding sources: None.

Background

Deep learning (DL) has received much attention as a solution for automatically diagnosing atrial fibrillation (AF) from raw ECG signals. However, few studies exist to investigate how DL approaches can be optimally configured and whether their diagnostic performance is externally validated.

Purpose

To explore how signal-related parameter tuning affects the ability of DL approaches to diagnose AF and validate the optimal approach internally and externally.

Methods

We applied two dedicated DL models (InceptionTime and MINIROCKET) on a set of 7,966 AF and non-AF (normal or with other abnormalities) ambulatory ECG samples, originating from the MIT-BIH AF, MIT-BIH Normal Sinus Rhythm and Long Term AF databases. We tested the effect of different sample lengths (30sec (s), 10s, 30/10s -30s with a "sliding window" of 10s-), sampling frequencies (200, 100, 50 Hz) and lead numbers (two-, single-), and the role of denoising (Discrete Wavelet Transformation, no denoising) on the ability to diagnose AF, by measuring ROC AUC and sensitivity (SEN) after repeated model training and testing. Under the optimal configuration, we trained 10 replicas of both models on 90% of the data and tested their performance on the remaining 10% (internal validation). Finally, we applied both pre-trained models on a separate dataset (MIT-BIH Arrhythmia) to determine their external validity.

Results

Although the diagnostic performance did not differ between 30s and 10s signals, the 30/10s setting displayed significantly higher median AUC (0.98) and sensitivity (97.3%, p&lt;0.05 for all comparisons). Signals sampled at 50Hz performed poorer (AUC=0.88, SEN=79.9%) than those at 100Hz (AUC=0.92, SEN=88.7%) and 200Hz (AUC=0.93, SEN=89.2%), although this difference slightly failed to reach statistical significance. Despite denoised signals showing a higher median AUC (0.95 vs. 0.92) and sensitivity (92.8% vs. 88.7%), the difference was not found significant. Similarly, two-lead signals performed better than single-lead ones (AUC=0.92 vs. 0.9 and SEN=88.7% vs. 84.1%, respectively), but without crossing the significance threshold. The internal validation with denoised, 30/10s, two-lead signals, at 100Hz, yielded similarly high performance metrics for both InceptionTime and MINIROCKET (AUC=0.98, SEN=96.9% and AUC=0.98, SEN=97.4%, respectively). In contrast, the performance on the external set dropped significantly (AUC=0.79, SEN=81.4% and AUC=0.72, SEN=83.7%, respectively, p&lt;0.001 for all comparisons).

Conclusions

Both DL approaches can effectively detect AF in ambulatory ECG signals, with only 3 out of 100 cases missed, designating their promising utility as screening tools for automated AF detection. While optimising tunable parameters can enhance the internal performance of such efforts, their external validation is necessary to establish their robustness "in the wild", since their performance on "unseen" data can be, similarly to our case, notably lower.

Mid-term endothelial dysfunction post COVID-19

Introduction

Cardiovascular complications of Coronavirus disease (COVID-19), resulting from the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), have been documented. Endothelium-induced “cytokine storm” in critically ill COVID-19 patients is one of the leading causes of morbidity and mortality. Vascular endothelial damage caused by COVID-19 emphasizes the crucial role of endothelium in COVID-19 clinical impact.

Purpose

To examine the mid-term (1-month) impact of COVID-19 in endothelial function.

Methods

In this case control study, 20 consecutive patients who were hospitalized for COVID-19 either on Intensive Care Unit (ICU) or non-ICU were examined one month following hospital discharge. In the control group we recruited 12 consecutive subjects from the outpatient cardiology clinic. Demographic and clinical data were collected, and endothelial function was evaluated by brachial artery flow-mediated dilation (FMD).

Results

There was no difference in age between COVID-19 patients and control subjects (66±12 years vs. 71±5 years, p&lt;0.17), in male sex (63% vs. 54%, p=0.66) in history of diabetes mellitus (27% vs. 36%, p=0.64), hypertension (36% vs. 54%, p=0.39), cardiovascular disease (27% vs.18%, p=0.61). From the COVID-19 subjects 65% were overweight or obese. During their hospitalization [3 ICU (15%)/17 non-ICU (85%), mean days: 17±6.7], 4 (20%) of COVID-19 patients developed ARDS, while single cases of stress-induced cardiomyopathy, pulmonary embolism, and acute coronary syndrome were detected. One month post discharge D-dimers (0.71±0.55 μg/ml) levels were above upper reference limit. Importantly, FMD one month after hospital discharge date, was significantly impaired in the COVID-19 group (3.59±1.63% vs. 9.31±2.98%, p&lt;0.001) compared to control group.

Conclusion

Post COVID-19 subjects one month post discharge have significant impaired endothelial function compared to control subjects. These findings highlight the significant interaction of COVID-19 with arterial endothelium and merit further research to conclude on the exact impact of vascular endothelium in physical history of SARS-CoV-2 infection.

Funding Acknowledgement

Type of funding sources: None.