Dr Eleni Elia

Spiking neural networks (SNNs) are a large class of neural model distinct from ‘classical’ continuous-valued networks such as multilayer perceptrons (MLPs). With event-driven dynamics and a continuous-time model in contrast to the discrete-time model of their classical counterparts, they offer interesting advantages in representational capacity and energy consumption. Spiking networks may also be more biologically plausible, offering more insights into neuroscience. However, developing models of learning for SNNs has historically proven challenging: as discrete-time systems, their dynamics are much more complex and they cannot benefit from the strong theoretical developments in MLPs such as convergence proofs and optimal gradient descent. Nor do they gain automatically from algorithmic improvements that
have produced efficient matrix inversion and batch training methods. Most of the existing research has focused on the most well-studied learning mechanism in SNNs, spike-timing-dependent plasticity (STDP), and although there has been progress, there are also notable pathologies that have often been
solved with a variety of ad-hoc techniques. While efforts have been made to map SNNs to classical convolutional neural networks (CNNs), these have not yet shown any decisive efficiency advantage over conventional CNNs. More promising research directions lie in the realm of pure spiking learning models
that exploit the inherent temporal dynamics (and often leverage recurrency). Metrics are needed; one possibility would be a measure of total energy cost per unit reduction in error. This tutorial overview looks at existing techniques for learning in SNNs and offers some thoughts for future directions.

Background: The right ventricle (RV) has complex geometry and function, with motion along three separate axes—longitudinal, radial, and anteroposterior. Quantitative assessment of RV function by two-dimension echocardiography (2DE) has been limited as a consequence of this complexity, whereas newer three dimensional (3D) analysis offers the potential for more comprehensive assessment of the contributors to RV function. The aims of this study were to quantify the longitudinal, radial and anteroposterior components of global RV function using 3D echocardiography in a cohort of healthy children and to examine maturational changes in these parameters.

Methods: Three-dimensional contours of the RV were generated from a cohort of healthy pediatric patients with structurally normal hearts at two centers. Traditional 2D and 3D echo characteristics were recorded. Using offline analysis of 3D datasets, RV motion was decomposed into three components, and ejection fractions (EF) were calculated (longitudinal-LEF; radial-REF; and anteroposterior-AEF). The individual decomposed EF values were indexed against the global RVEF. Strain values were calculated as well.

Results: Data from 166 subjects were included in the analysis; median age was 13.5 years (range 0 to 17.4 years). Overall, AEF was greater than REF and LEF (29.2 ± 6.2% vs. 25.1 ± 7.2% and 25.7 ± 6.0%, respectively; p 

Conclusions: In healthy children, anteroposterior shortening is the dominant component of RV contraction. Evaluation of 3D parameters of the RV in children is feasible and enhances the overall understanding of RV function, which may allow improvements in recognition of dysfunction and assessment of treatment effects in the future.

BACKGROUND. In pediatric echocardiography, reference intervals are required to distinguish normal variation from pathology. Left ventricular (LV) parameters are particularly important predictors of clinical outcome. However, data from healthy newborns are limited, and current reference intervals provide an inadequate approximation of normal reference ranges.

OBJECTIVES. Normative reference intervals and z-scores for 2-dimensional echocardiographic measurements of LV structure and function based on a large group of healthy newborns were developed.

METHODS. The study population included 13,454 healthy newborns from the Copenhagen Baby Heart Study who were born at term to healthy mothers, had an echocardiogram performed within 30 days of birth, and did not have congenital heart disease. To develop normative reference intervals, this study modeled 10 LV parameters as a function of body surface area through joint modeling of 4 statistical components.

RESULTS. Infants in the study population (48.5% were female) had a median body surface area of 0.23 m2 (IQR: 0.22-0.25 m2) and median age of 12.0 days (IQR: 8.0-15.0 days) at examination. All normative reference intervals performed well in both sexes without stratification on infant sex. In contrast, creation of separate reference models for infants examined at

CONCLUSIONS. This study provides normative reference intervals and z-scores for 10 clinical, widely used echocardiographic measures of LV structure and function based on a large cohort of newborns. These results provide highly needed reference material for clinical application by pediatric cardiologists.

Purpose. 
We sought to establish whether nucleated red blood cells (NRBCs) are predictive of disposition, morbidity, and mortality for pediatric patients presenting to the emergency department (ED).

Methods.  
A single-center retrospective cohort study examining all ED encounters from patients aged younger than 19 years between January 2016 and March 2020, during which a complete blood count was obtained. Univariate analysis and multivariable logistic regression were used to test the presence of NRBCs as an independent predictor of patient-related outcomes.

Results. 
The prevalence of NRBCs was 8.9% (4195/46,991 patient encounters). Patient with NRBCs were younger (median age 4.58 vs 8.23 years; P

Conclusions. 
The presence of NRBCs is an independent predictor for mortality, including in-hospital mortality, ICU admission, CPR, and readmission within 30 days for children presenting to the ED.

Objective: To characterize distinct comorbidities, outcomes, and treatment patterns in children with Down syndrome and pulmonary hypertension in a large, multicenter pediatric pulmonary hypertension registry.

Study design: We analyzed data from the Pediatric Pulmonary Hypertension Network (PPHNet) Registry, comparing demographic and clinical characteristics of children with Down syndrome and children without Down syndrome. We examined factors associated with pulmonary hypertension resolution and a composite outcome of pulmonary hypertension severity in the cohort with Down syndrome.

Results: Of 1475 pediatric patients with pulmonary hypertension, 158 (11%) had Down syndrome. The median age at diagnosis of pulmonary hypertension in patients with Down syndrome was 0.49 year (IQR, 0.21-1.77 years), similar to that in patients without Down syndrome. There was no difference in rates of cardiac catheterization and prescribed pulmonary hypertension medications in children with Down syndrome and those without Down syndrome. Comorbidities in Down syndrome included congenital heart disease (95%; repaired in 68%), sleep apnea (56%), prematurity (49%), recurrent respiratory exacerbations (35%), gastroesophageal reflux (38%), and aspiration (31%). Pulmonary hypertension resolved in 43% after 3 years, associated with a diagnosis of pulmonary hypertension at age

Conclusions: Despite high rates of cardiac and respiratory comorbidities that influence the severity of pulmonary hypertension, children with Down syndrome-associated pulmonary hypertension generally have a survival rate similar to that of children with non-Down syndrome-associated pulmonary hypertension. Resolution of pulmonary hypertension is common but reduced in children with complicated respiratory comorbidities

While care models adapt to the COVID-19 pandemic with virtual and hybrid visits, clinical factors associated with treatment changes among ambulatory pediatric pulmonary arterial hypertension (PAH) patients are not well characterized. To understand which data critically altered treatment recommendations, we conducted a retrospective review among ambulatory children with Group 1 PAH to determine optimal visit and diagnostic strategies. Changes in management included: unplanned new treatments, dose modifications of vasodilators or diuretics, unscheduled hospitalizations, or changes to activity recommendations, catheterization schedule, or other testing. Factors prompting management changes were classified as symptoms, exam findings, or diagnostic tests. Across 398 ambulatory visits by 48 patients, 38 patients (79%) at 88 visits (22%) required change in management, most commonly in targeted PH medication. Changes were driven by symptoms alone (15%), diagnostic testing alone (47%), exam only (2%), symptoms and exam (2%), combination of testing and symptoms or testing and exam (25%), and other reasons (9%). Patients with World Health Organization functional Class IV (odds ratio [OR] 9.04 vs. Class I, p = 0.014) or Class III (OR 2.08 vs. Class I, p = 0.050) were more likely to undergo change in management. However, among Class I patients, 18% of visits generated changes in management because of test findings. While multiple factors affect management in ambulatory pediatric PH, neither symptoms nor exam was sufficient for identifying patients warranting clinical change in management. Testing accounted for most changes. Thus, in-person or hybrid surveillance including history, exam, and diagnostic testing remains essential for optimal management of pediatric PAH.

Importance: Elective surgeries are primarily scheduled according to surgeon availability with less consideration of patients' postoperative cardiac intensive care unit (CICU) length of stay. Furthermore, the CICU census can exhibit a high rate of variation in which the CICU is operating at over-capacity, resulting in admission delays and cancellations; or under-capacity, resulting in underutilized labor and overhead expenditures.

Objective: To identify strategies to reduce variation in CICU occupancy levels and avoid late patient surgery cancellation.

Design: Monte Carlo simulation study of the daily and weekly CICU census at Boston Children's Hospital Heart Center. Data on all surgical admissions to and discharges from the CICU at Boston Children's Hospital between September 1, 2009 and November 2019 were included to obtain the distribution of length of stay for the simulation study. The available data allows us to model realistic length of stay samples that include short and extended lengths of stay.

Main Outcomes: Annual number of patient surgical cancellations and change in average daily census.

Results: We demonstrate that the models of strategic scheduling would result in up to 57% reduction in patient surgical cancellations, increase the historically low Monday census and decrease the historically higher late-mid-week (Wednesday and Thursday) censuses in our center.

Conclusions and Relevance: Use of strategic scheduling may improve surgical capacity and reduce the number of annual cancellations. The reduction of peaks and valleys in the weekly census corresponds to a reduction of underutilization and overutilization of the system.

Background: Biomarkers are increasingly part of assessing and managing heart failure (HF) in adults with congenital heart disease (CHD).

Objectives: To understand the response of cardiac biomarkers with therapy for acute decompensated heart failure (ADHF) and the relationship to prognosis after discharge in adults with CHD.

Design: A prospective, observational cohort study with serial blood biomarker measurements.

Settings: Single-center study in the inpatient setting with outpatient follow-up.

Participants: Adults (≥18 years old) with CHD admitted with ADHF between August 1, 2019, and March 1, 2020.

Exposure: We measured body mass, Kansas City Cardiomyopathy Questionnaire (KCCQ-12) score, N-terminal pro-B-type natriuretic peptide (NT-proBNP), and high-sensitivity C-reactive protein (hsCRP) at enrollment, discharge, and 1st clinic follow-up visit; soluble suppression of tumorigenicity 2 (sST2) was measured at the first two time points.

Measures: Univariate regression assessed the association between changes in weight, biomarkers, and changes in KCCQ-12 scores, between enrollment and discharge (Δ Hospitalization ) and between discharge and 1st clinical follow-up visit (Δ Post-discharge ). Wilcoxon rank-sum tests assessed the association between change in biomarkers, KCCQ-12 scores, and the composite outcome of cardiovascular death or rehospitalization for ADHF.

Results: A total of 26 patients were enrolled. The median age was 51.9 years [IQR: 38.8, 61.2], 13 (54.2%) were women, and median hospital stay was 6.5 days [IQR: 4.0, 15.0] with an associated weight loss of 2.8 kg [IQR -5.1, -1.7]. All three cardiac biomarkers decreased during hospitalization with diuresis while KCCQ-12 scores improved; a greater decrease in sST2 was associated with an improved KCCQ-12 symptom frequency (SF) subdomain score (p = 0.012), but otherwise, there was no significant relationship between biomarkers and KCCQ-12 change. Change in hsCRP and NT-proBNP after discharge was not associated with the composite outcome (n = 8, vs. n = 16 who did not experience the outcome; Δ Post-discharge hsCRP +5.1 vs. -1.0 mg/l, p = 0.061; NT-proBNP +785.0 vs. +130.0 pg/ml, p = 0.220).

Conclusions: Serial biomarker measurements respond to acute diuresis in adults with CHD hospitalized for ADHF. These results should motivate further research into the use of biomarkers to inform HF therapy in adults with CHD.

Background: Although survival of children with advanced heart disease (AHD) has been improving, long-term morbidity remains high, leading to chronic conditions. There is little data on the perspectives of fathers caring for chronically ill children. This study describes the experience and prognostic awareness of fathers of hospitalized children with AHD.

Methods: Cross-sectional survey study of parents caring for children hospitalized with AHD (age 30 d - 19 yr) admitted for ≥ 7 d between 2/18-2/19. Inclusion criteria: single ventricle, pulmonary vein stenosis, pulmonary hypertension, length of stay >30 d, mechanical support, mechanical ventilation >14 d, or ≥3 admissions/yr. One parent per patient completed surveys, which included mothers(128) and fathers(27). Data from the fathers’ surveys were analyzed using descriptive methods.

Results: The 27 fathers enrolled at a median age of 34 yrs (range, 22 - 54) and 20/27 were married, with 70% identified as white. Median patient age was 2 yrs (IQR, 0-4) and 37% were in the CICU at time of survey. Diagnoses included single ventricle(10), pulmonary hypertension(5), other congenital(11), and transplant(1). All but one (96%) of the fathers surveyed thought they understood their child’s prognosis “extremely well” or “well,” and 59% thought that in the past week they were “very prepared” for the medical problems their child was experiencing. However, 56% of fathers wanted to know more about prognosis and 22% thought their child’s care team knew something about their child’s prognosis that they did not know. A little over half (52%) of fathers reported feeling unsure if their child was getting better or worse and 29% of fathers thought they received conflicting information from the healthcare team. Lifelong treatment for their child was felt to be necessary by 11 fathers, with physical activity restrictions noted by 7 and learning/development/behavior restrictions by 8. One third of fathers (35%) reported they had to tell health professionals medical details of their child’s condition because the health professionals did not seem to know. Of the 13 fathers who reported they had a conversation with their child’s healthcare team that included discussion of what would happen if their child got sicker, every father found this conversation helpful, although 9 (69%) reported it was “somewhat” or “a little” confusing. Nearly all felt that the conversation both allowed them to express their worries (95%) and that their worries were adequately addressed (95%).

Conclusion: Fathers report that they understand prognosis well, however, only half feel very prepared for the medical problems their child is experiencing. They feel supported in their worries, however many fathers consider lifelong limitations to be unlikely. Future research that explores potential differences in mothers and fathers will be beneficial in order to address specific educational needs of fathers.

Background: Progression free survival (PFS) and tumour response (TR) have been investigated as surrogate endpoints for overall survival (OS) in advanced colorectal cancer (aCRC), however their validity has been shown to be suboptimal. In recent years, meta-analytic methods allowing for use of multiple surrogate endpoints jointly have been proposed. Our aim was to assess if PFS and TR used jointly as surrogate endpoints to OS improve their predictive value.

Methods: Data were obtained from a systematic review of randomised controlled trials investigating effectiveness of pharmacological therapies in aCRC, including systemic chemotherapies, anti-epidermal growth factor receptor therapies and anti-angiogenic agents. Multivariate meta-analysis was used to model the association patterns between treatment effects on the surrogate endpoints (TR, PFS) and the final outcome (OS).

Results: Analysis of 33 trials reporting treatment effects on all three outcomes showed reasonably strong association between treatment effects on PFS and OS, however the association parameters were obtained with a large uncertainty. A weak surrogate relationship was noted between the treatment effects on TR and OS. Modelling the two surrogate endpoints, TR and PFS, jointly as predictors of treatment effect on OS gave no marked improvement to surrogate association patterns. Modest improvement in the precision of the predicted treatment effects on the final outcome was noted in studies investigating anti-angiogenic therapy, however it was likely due to chance.

Conclusion: The joint use of two surrogate endpoints did not lead to marked improvement in the association between treatment effects on surrogate and final endpoints in advanced colorectal cancer.

Introduction: The aim of this study was to determine the prognostic value of the first urinary albumin/creatinine ratio (ACR) for adverse maternal and neonatal outcomes and how it relates to other prognostic factors.

Material and methods: We performed a retrospective cohort study from December 2009 to February 2012 with analysis of demographic, clinical and biochemical data from two obstetric day assessment units in hospitals in Southeast Scotland. We included 717 pregnant women, with singleton pregnancies after 20 weeks' gestation, referred for evaluation of suspected preeclampsia and having their first ACR performed. The ability of ACR to predict future outcomes was assessed in both univariable and multivariable logistic regression models. The latter assessed its prognostic value independent of (adjusting for) existing prognostic factors. Primary outcome measures were maternal and neonatal composite adverse outcomes, and a secondary outcome was gestation at delivery.

Results: In all, 204 women (28.5%) experienced a composite adverse maternal outcome and 146 women (20.4%) experienced a composite adverse neonatal outcome. Multivariate analysis of log-transformed ACR demonstrated that a 1-unit increase in log ACR is associated with an increased odds of adverse maternal [odds ratio 1.60, 95% confidence interval (CI) 1.45-1.80] and adverse neonatal (odds ratio 1.15, 95% CI 1.02-1.29) composite outcomes, and with reduced gestational age at delivery (coefficient: -0.46, 95% CI -0.54 to -0.38).

Conclusions: ACR is an independent prognostic factor for maternal and neonatal adverse outcomes in suspected preeclampsia. ACR may be useful to inform risk predictions within a prognostic model.

A prognostic factor is any measure that is associated with the risk of future health outcomes in those with existing disease. Often, the prognostic ability of a factor is evaluated in multiple studies. However, meta-analysis is difficult because primary studies often use different methods of measurement and/or different cut-points to dichotomise continuous factors into 'high' and 'low' groups; selective reporting is also common. We illustrate how multivariate random effects meta-analysis models can accommodate multiple prognostic effect estimates from the same study, relating to multiple cut-points and/or methods of measurement. The models account for within-study and between-study correlations, which utilises more information and reduces the impact of unreported cut-points and/or measurement methods in some studies. The applicability of the approach is improved with individual participant data and by assuming a functional relationship between prognostic effect and cut-point to reduce the number of unknown parameters. The models provide important inferential results for each cut-point and method of measurement, including the summary prognostic effect, the between-study variance and a 95% prediction interval for the prognostic effect in new populations. Two applications are presented. The first reveals that, in a multivariate meta-analysis using published results, the Apgar score is prognostic of neonatal mortality but effect sizes are smaller at most cut-points than previously thought. In the second, a multivariate meta-analysis of two methods of measurement provides weak evidence that microvessel density is prognostic of mortality in lung cancer, even when individual participant data are available so that a continuous prognostic trend is examined (rather than cut-points).