Deep learning-based model detects atrial septal defects from electrocardiography: a cross-sectional multicenter hospital-based study.

Background: Atrial septal defect (ASD) increases the risk of adverse cardiovascular outcomes. Despite the potential for risk mitigation through minimally invasive percutaneous closure, ASD remains underdiagnosed due to subtle symptoms and examination findings. To bridge this diagnostic gap, we propose a novel screening strategy aimed at early detection and enhanced diagnosis through the implementation of a convolutional neural network (CNN) to identify ASD from 12-lead electrocardiography (ECG). Methods: ECGs were collected from patients with at least one recorded echocardiogram at 3 hospitals from 2 continents (Keio University Hospital from July 2011 to December 2020, Brigham and Women's Hospital from January 2015 to December 2020, and Dokkyo Medical University Saitama Medical Center from January 2010 and December 2021). ECGs from patients with a diagnosis of ASD were labeled as positive cases while the remainder were labeled as negative. ECGs after the closure of ASD were excluded. After randomly splitting the ECGs into 3 datasets (50% derivation, 20% validation, and 30% test) with no patient overlap, a CNN-based model was trained using the derivation datasets from 2 hospitals and was tested on held-out datasets along with an external validation on the 3rd hospital. All eligible ECGs were used for derivation and validation whereas the earliest ECG for each patient was used for the test and external validation. The discrimination of ASD was assessed by the area under the receiver operating characteristic curve (AUROC). Multiple subgroups were examined to identify any heterogeneity. Findings: A total of 671,201 ECGs from 80,947 patients were collected from the 3 institutions. The AUROC for detecting ASD was 0.85-0.90 across the 3 hospitals. The subgroup analysis showed excellent performance across various characteristics Screening simulation using the model greatly increased sensitivity from 80.6% to 93.7% at specificity 33.6% when compared to using overt ECG abnormalities. Interpretation: A CNN-based model using 12-lead ECG successfully identified the presence of ASD with excellent generalizability across institutions from 2 separate continents. Funding: This work was supported by research grants from JST (JPMJPF2101), JSR corporation, Taiju Life Social Welfare Foundation, Kondou Kinen Medical Foundation, Research fund of Mitsukoshi health and welfare foundation, Tokai University School of Medicine Project Research and Internal Medicine Project Research, Secom Science and Technology Foundation, and Grants from AMED (JP23hma922012 and JP23ym0126813). This work was partially supported by One Brave Idea, co-funded by the American Heart Association and Verily with significant support from AstraZeneca and pillar support from Quest Diagnostics.

Cardiovascular Risk Assessment Using Artificial Intelligence-Enabled Event Adjudication and Hematologic Predictors.

BACKGROUND: Researchers routinely evaluate novel biomarkers for incorporation into clinical risk models, weighing tradeoffs between cost, availability, and ease of deployment. For risk assessment in population health initiatives, ideal inputs would be those already available for most patients. We hypothesized that common hematologic markers (eg, hematocrit), available in an outpatient complete blood count without differential, would be useful to develop risk models for cardiovascular events. METHODS: We developed Cox proportional hazards models for predicting heart attack, ischemic stroke, heart failure hospitalization, revascularization, and all-cause mortality. For predictors, we used 10 hematologic indices (eg, hematocrit) from routine laboratory measurements, collected March 2016 to May 2017 along with demographic data and diagnostic codes. As outcomes, we used neural network-based automated event adjudication of 1 028 294 discharge summaries. We trained models on 23 238 patients from one hospital in Boston and evaluated them on 29 671 patients from a second one. We assessed calibration using Brier score and discrimination using Harrell's concordance index. In addition, to determine the utility of high-dimensional interactions, we compared our proportional hazards models to random survival forest models. RESULTS: Event rates in our cohort ranged from 0.0067 to 0.075 per person-year. Models using only hematology indices had concordance index ranging from 0.60 to 0.80 on an external validation set and showed the best discrimination when predicting heart failure (0.80 [95% CI, 0.79-0.82]) and all-cause mortality (0.78 [0.77-0.80]). Compared with models trained only on demographic data and diagnostic codes, models that also used hematology indices had better discrimination and calibration. The concordance index of the resulting models ranged from 0.75 to 0.85 and the improvement in concordance index ranged up to 0.072. Random survival forests had minimal improvement over proportional hazards models. CONCLUSIONS: We conclude that low-cost, ubiquitous inputs, if biologically informative, can provide population-level readouts of risk.

Ecosystem Barriers to Innovation Adoption in Clinical Practice.

Despite increasing ability to understand and correct molecular derangements in disease, genomics and novel phenotypic assays are unevenly deployed in clinical practice. This has hampered translational research and our ability to identify clinically actionable subtypes of disease. Historic examples illustrate how the perspectives of stakeholders across the healthcare ecosystem can influence adoption of innovations in healthcare. Consideration of these factors, from discovery to implementation, can accelerate adoption of new molecular and digital phenotypes in a 'learning' healthcare ecosystem.

Development of an entirely remote, non-physician led hypertension management program.

BACKGROUND: Hypertension remains poorly controlled on the population level. National rates of control, even when defined leniently by BP < 140/90 mm Hg, are only ~50%. As growing healthcare costs coincide with tighter blood pressure (BP) targets, innovative management programs are needed to maximize efficiency of care delivery and optimize control. HYPOTHESIS: We aimed to develop a remote, navigator-led hypertension innovation program that would leverage algorithmic care pathways, home BP measurements and patient coaching to allow rapid and complete medication titration. METHODS: A multidisciplinary group of clinical experts from subspecialties and primary care collaborated to develop an evidence-based clinical algorithm, designed to be automated and administered by non-licensed patient navigators. In the development stage, a prospective pilot cohort of 130 patients was managed by nurse practitioners and pharmacists to ensure efficacy and safety. Patients with clinic BP ≥ 140/90 mm Hg were enrolled and given a Bluetooth-enabled BP device. Home BPs were transmitted automatically into the electronic medical record. Medication titrations were performed by phone at biweekly intervals, based upon weekly average BP, until home BP was controlled at <135/85 mm Hg. RESULTS: Eighty-one percent of all enrolled, and 91% of those patients who regularly measured home BP achieved goal, in an average of 7 weeks. Control was reached similarly across races, genders, and ages. CONCLUSIONS: A home-based BP control program run by non-physicians can provide efficient, effective and rapid control, suggesting an innovative paradigm for hypertension management. This program is effective, sustainable, adaptable, and scalable to fit current and emerging national systems of healthcare.

The Impact of Whole-Genome Sequencing on the Primary Care and Outcomes of Healthy Adult Patients: A Pilot Randomized Trial.

Background: Whole-genome sequencing (WGS) in asymptomatic adults might prevent disease but increase health care use without clinical value. Objective: To describe the effect on clinical care and outcomes of adding WGS to standardized family history assessment in primary care. Design: Pilot randomized trial. (ClinicalTrials.gov: NCT01736566). Setting: Academic primary care practices. Participants: 9 primary care physicians (PCPs) and 100 generally healthy patients recruited at ages 40 to 65 years. Intervention: Patients were randomly assigned to receive a family history report alone (FH group) or in combination with an interpreted WGS report (FH + WGS group), which included monogenic disease risk (MDR) results (associated with Mendelian disorders), carrier variants, pharmacogenomic associations, and polygenic risk estimates for cardiometabolic traits. Each patient met with his or her PCP to discuss the report. Measurements: Clinical outcomes and health care use through 6 months were obtained from medical records and audio-recorded discussions between PCPs and patients. Patients' health behavior changes were surveyed 6 months after receiving results. A panel of clinician-geneticists rated the appropriateness of how PCPs managed MDR results. Results: Mean age was 55 years; 58% of patients were female. Eleven FH + WGS patients (22% [95% CI, 12% to 36%]) had new MDR results. Only 2 (4% [CI, 0.01% to 15%]) had evidence of the phenotypes predicted by an MDR result (fundus albipunctatus due to RDH5 and variegate porphyria due to PPOX). Primary care physicians recommended new clinical actions for 16% (CI, 8% to 30%) of FH patients and 34% (CI, 22% to 49%) of FH + WGS patients. Thirty percent (CI, 17% to 45%) and 41% (CI, 27% to 56%) of FH and FH + WGS patients, respectively, reported making a health behavior change after 6 months. Geneticists rated PCP management of 8 MDR results (73% [CI, 39% to 99%]) as appropriate and 2 results (18% [CI, 3% to 52%]) as inappropriate. Limitation: Limited sample size and ancestral and socioeconomic diversity. Conclusion: Adding WGS to primary care reveals new molecular findings of uncertain clinical utility. Nongeneticist providers may be able to manage WGS results appropriately, but WGS may prompt additional clinical actions of unclear value. Primary Funding Source: National Institutes of Health.

Future translational applications from the contemporary genomics era: a scientific statement from the American Heart Association.

The field of genetics and genomics has advanced considerably with the achievement of recent milestones encompassing the identification of many loci for cardiovascular disease and variable drug responses. Despite this achievement, a gap exists in the understanding and advancement to meaningful translation that directly affects disease prevention and clinical care. The purpose of this scientific statement is to address the gap between genetic discoveries and their practical application to cardiovascular clinical care. In brief, this scientific statement assesses the current timeline for effective translation of basic discoveries to clinical advances, highlighting past successes. Current discoveries in the area of genetics and genomics are covered next, followed by future expectations, tools, and competencies for achieving the goal of improving clinical care.

Heart on a plate: histological and functional assessment of isolated adult zebrafish hearts maintained in culture.

The zebrafish is increasingly used for cardiovascular genetic and functional studies. We present a novel protocol to maintain and monitor whole isolated beating adult zebrafish hearts in culture for long-term experiments. Excised whole adult zebrafish hearts were transferred directly into culture dishes containing optimized L-15 Leibovitz growth medium and maintained for 5 days. Hearts were assessed daily using video-edge analysis of ventricle function using low power microscopy images. High-throughput histology techniques were used to assess changes in myocardial architecture and cell viability. Mean spontaneous Heart rate (HR, min(-1)) declined significantly between day 0 and day 1 in culture (96.7 ± 19.5 to 45.2 ± 8.2 min-1, mean ± SD, p = 0.001), and thereafter declined more slowly to 27.6 ± 7.2 min(-1) on day 5. Ventricle wall motion amplitude (WMA) did not change until day 4 in culture (day 0, 46.7 ± 13.0 µm vs day 4, 16.9 ± 1.9 µm, p = 0.08). Contraction velocity (CV) declined between day 0 and day 3 (35.6 ± 14.8 vs 15.2 ± 5.3 µms(-1), respectively, p = 0.012) while relaxation velocity (RV) declined quite rapidly (day 0, 72.5 ± 11.9 vs day 1, 29.5 ± 5.8 µms(-1), p = 0.03). HR and WMA responded consistently to isoproterenol from day 0 to day 5 in culture while CV and RV showed less consistent responses to beta-agonist. Cellular architecture and cross-striation pattern of cardiomyocytes remained unchanged up to day 3 in culture and thereafter showed significant deterioration with loss of striation pattern, pyknotic nuclei and cell swelling. Apoptotic markers within the myocardium became increasingly frequent by day 3 in culture. Whole adult zebrafish hearts can be maintained in culture-medium for up to 3 days. However, after day-3 there is significant deterioration in ventricle function and heart rate accompanied by significant histological changes consistent with cell death and loss of cardiomyocyte cell integrity. Further studies are needed to assess whether this preparation can be optimised for longer term survival.

Techniques for the in vivo assessment of cardio-renal function in zebrafish (Danio rerio) larvae.

Zebrafish, a well-established vertebrate model, offer unique advantages for assessing renal function and physiology. Assays determining renal glomerular function based on cardiovascular erythrocyte flow and reduction of injected FITC-inulin were developed, each validated using the nephrotoxin gentamicin. Bland­Atlman analysis showed a strong association between measurements of the rate of inulin excretion and that of fluorescent reduction from the arterial vasculature. Reduced renal clearance of inulin, resulting from gentamicin or NaCl loading, was concurrent with reduced erythrocyte velocity, and yolk sac and pericardium oedema. These techniques, assessing pronephric function, highlight the potential for in vivo physiological study in this genetically tractable model.

In vivo recording of adult zebrafish electrocardiogram and assessment of drug-induced QT prolongation.

In the last decade the zebrafish has become a major model organism for the study of development and organogenesis. To maximize the experimental utility of this organism, it will be important to establish methods for adult phenotyping. We previously proposed that the embryonic zebrafish may be useful in high-throughput screening for drug-induced cardiotoxicity. We now describe a method for the reproducible recording of the adult zebrafish ECG and illustrate its application in the investigation of QT-prolonging drugs. Zebrafish ECGs were obtained by inserting two needle electrodes through the ventral epidermis. Fish were perfused orally, and motion artifacts were eliminated with a paralytic dose of mu-conotoxin GIIIB. Test compounds were delivered via the perfusion system. Without a means of hydration and oxygenation, the fish succumb rapidly. The use of a perfusion system allowed stable recording for > 6 h. Baseline conduction intervals were as follows: PR, 66 ms (SD 14); QRS, 34 ms (SD 11); QT, 242 ms (SD 54); and R-R, 398 ms (SD 77). The known QT-prolonging agents astemizole, haloperidol, pimozide, and terfenadine caused corrected QT increases of 18% (SD 9), 16% (SD 11), 17% (SD 9), and 11% (SD 6), respectively. The control drugs clonidine, penicillin and propranolol did not prolong the corrected QT interval. In conclusion, perfusion and muscular paralysis allows stable, low-noise recording of zebrafish ECGs. Agents known to cause QT prolongation in humans caused QT prolongation in fish in each case. The development of rigorous tools for the phenotyping of adult zebrafish will complement the high-throughput assays currently under development for embryonic and larval fish.