Search

Abstract Congenital long QT syndrome (LQTS) diagnosis is complicated by limited genetic testing at scale, low prevalence, and normal QT corrected interval in patients with high-risk genotypes. We developed a deep learning approach combining electrocardiogram (ECG) waveform and electronic health record data to assess whether patients had pathogenic variants causing LQTS. We defined patients with high-risk genotypes as having ≥1 pathogenic variant in one of the LQTS-susceptibility genes. We trained the model using data from United Kingdom Biobank (UKBB) and then fine-tuned in a racially/ethnically diverse cohort using Mount Sinai BioMe Biobank. Following group-stratified 5-fold splitting, the fine-tuned model achieved area under the precision-recall curve of 0.29 (95% confidence interval [CI] 0.28–0.29) and area under the receiver operating curve of 0.83 (0.82–0.83) on independent testing data from BioMe. Multimodal fusion learning has promise to identify individuals with pathogenic genetic mutations to enable patient prioritization for further work up.

Abstract Background Advances in self-supervised learning (SSL) have enabled state-of-the-art automated medical image diagnosis from small, labeled datasets. This label efficiency is often desirable, given the difficulty of obtaining expert labels for medical image recognition tasks. However, most efforts toward SSL in medical imaging are not adapted to video-based modalities, such as echocardiography. Methods We developed a self-supervised contrastive learning approach, EchoCLR, for echocardiogram videos with the goal of learning strong representations for efficient fine-tuning on downstream cardiac disease diagnosis. EchoCLR pretraining involves (i) contrastive learning, where the model is trained to identify distinct videos of the same patient, and (ii) frame reordering, where the model is trained to predict the correct of video frames after being randomly shuffled. Results When fine-tuned on small portions of labeled data, EchoCLR pretraining significantly improves classification performance for left ventricular hypertrophy (LVH) and aortic stenosis (AS) over other transfer learning and SSL approaches across internal and external test sets. When fine-tuning on 10% of available training data (519 studies), an EchoCLR-pretrained model achieves 0.72 AUROC (95% CI: [0.69, 0.75]) on LVH classification, compared to 0.61 AUROC (95% CI: [0.57, 0.64]) with a standard transfer learning approach. Similarly, using 1% of available training data (53 studies), EchoCLR pretraining achieves 0.82 AUROC (95% CI: [0.79, 0.84]) on severe AS classification, compared to 0.61 AUROC (95% CI: [0.58, 0.65]) with transfer learning. Conclusions EchoCLR is unique in its ability to learn representations of echocardiogram videos and demonstrates that SSL can enable label-efficient disease classification from small amounts of labeled data.

Abstract Acute myeloid leukemia (AML) is fatal in the majority of adults. Identification of new therapeutic targets and their pharmacologic modulators are needed to improve outcomes. Previous studies had shown that immunization of rabbits with normal peripheral WBCs that had been incubated with fluorodinitrobenzene elicited high titer antibodies that bound to a spectrum of human leukemias. We report that proteomic analyses of immunoaffinity-purified lysates of primary AML cells showed enrichment of scaffolding protein IQGAP1. Immunohistochemistry and gene-expression analyses confirmed IQGAP1 mRNA overexpression in various cytogenetic subtypes of primary human AML compared to normal hematopoietic cells. shRNA knockdown of IQGAP1 blocked proliferation and clonogenicity of human leukemia cell-lines. To develop small molecules targeting IQGAP1 we performed in-silico screening of 212,966 compounds, selected 4 hits targeting the IQGAP1-GRD domain, and conducted SAR of the ‘fittest hit’ to identify UR778Br, a prototypical agent targeting IQGAP1. UR778Br inhibited proliferation, induced apoptosis, resulted in G2/M arrest, and inhibited colony formation by leukemia cell-lines and primary-AML while sparing normal marrow cells. UR778Br exhibited favorable ADME/T profiles and drug-likeness to treat AML. In summary, AML shows response to IQGAP1 inhibition, and UR778Br, identified through in-silico studies, selectively targeted AML cells while sparing normal marrow.

Abstract This research investigates the impact of storage conditions on the quality and preservation of 'Shalimar' apples, a relatively new cultivar known for its resistance to apple scab and powdery mildew. The study explores the efficacy of different storage techniques such as regular atmosphere (RA), controlled atmosphere (CA), and dynamic controlled atmosphere with CO2 Monitoring (DCA-CD), as well as the integration of 1-methylcyclopropene (1-MCP) at different storage temperatures (1 °C and 3 °C). Various fruit quality parameters were monitored under different storage conditions, including firmness, titratable acidity, total soluble solids, background color, respiration, ethylene production, and volatile compounds. The results indicate that the controlled atmosphere (CA) at 1 °C emerges as an efficient method for long-term storage. However, it is noted that CA storage may impact the apple aroma, emphasizing the need for a balance between preservation and consumer acceptability. On the other hand, DCA-CD at variable temperatures (approximately 2.5 °C) offers a promising approach for maintaining fruit quality and a higher concentration of volatile compounds. Integrating 1-MCP enhances firmness, but its impact varies across storage conditions. Principal component analysis (PCA) provides insights into the relationships between storage conditions, fruit quality, and volatile compounds. This study contributes valuable insights into optimizing storage strategies for ‘Shalimar’ apples, addressing sustainability and quality preservation in apple production.

Abstract SARS-CoV-2 reinfections increased substantially after Omicron variants emerged. Large-scale community-based comparisons across multiple Omicron waves of reinfection characteristics, risk factors, and protection afforded by previous infection and vaccination, are limited. Here we studied ~45,000 reinfections from the UK’s national COVID-19 Infection Survey and quantified the risk of reinfection in multiple waves, including those driven by BA.1, BA.2, BA.4/5, and BQ.1/CH.1.1/XBB.1.5 variants. Reinfections were associated with lower viral load and lower percentages of self-reporting symptoms compared with first infections. Across multiple Omicron waves, estimated protection against reinfection was significantly higher in those previously infected with more recent than earlier variants, even at the same time from previous infection. Estimated protection against Omicron reinfections decreased over time from the most recent infection if this was the previous or penultimate variant (generally within the preceding year). Those 14–180 days after receiving their most recent vaccination had a lower risk of reinfection than those >180 days from their most recent vaccination. Reinfection risk was independently higher in those aged 30–45 years, and with either low or high viral load in their most recent previous infection. Overall, the risk of Omicron reinfection is high, but with lower severity than first infections; both viral evolution and waning immunity are independently associated with reinfection.

Abstract Randomized clinical trials (RCT) represent the cornerstone of evidence-based medicine but are resource-intensive. We propose and evaluate a machine learning (ML) strategy of adaptive predictive enrichment through computational trial phenomaps to optimize RCT enrollment. In simulated group sequential analyses of two large cardiovascular outcomes RCTs of (1) a therapeutic drug (pioglitazone versus placebo; Insulin Resistance Intervention after Stroke (IRIS) trial), and (2) a disease management strategy (intensive versus standard systolic blood pressure reduction in the Systolic Blood Pressure Intervention Trial (SPRINT)), we constructed dynamic phenotypic representations to infer response profiles during interim analyses and examined their association with study outcomes. Across three interim timepoints, our strategy learned dynamic phenotypic signatures predictive of individualized cardiovascular benefit. By conditioning a prospective candidate’s probability of enrollment on their predicted benefit, we estimate that our approach would have enabled a reduction in the final trial size across ten simulations (IRIS: −14.8% ± 3.1%, p one-sample t-test = 0.001; SPRINT: −17.6% ± 3.6%, p one-sample t-test

Abstract Arabs account for 5% of the world population and have a high burden of cardiometabolic disease, yet clinical utility of polygenic risk prediction in Arabs remains understudied. Among 5399 Arab patients, we optimize polygenic scores for 10 cardiometabolic traits, achieving a performance that is better than published scores and on par with performance in European-ancestry individuals. Odds ratio per standard deviation (OR per SD) for a type 2 diabetes score was 1.83 (95% CI 1.74–1.92), and each SD of body mass index (BMI) score was associated with 1.18 kg/m2 difference in BMI. Polygenic scores associated with disease independent of conventional risk factors, and also associated with disease severity—OR per SD for coronary artery disease (CAD) was 1.78 (95% CI 1.66–1.90) for three-vessel CAD and 1.41 (95% CI 1.29–1.53) for one-vessel CAD. We propose a pragmatic framework leveraging public data as one way to advance equitable clinical implementation of polygenic scores in non-European populations.

Abstract Artificial intelligence (AI) can detect left ventricular systolic dysfunction (LVSD) from electrocardiograms (ECGs). Wearable devices could allow for broad AI-based screening but frequently obtain noisy ECGs. We report a novel strategy that automates the detection of hidden cardiovascular diseases, such as LVSD, adapted for noisy single-lead ECGs obtained on wearable and portable devices. We use 385,601 ECGs for development of a standard and noise-adapted model. For the noise-adapted model, ECGs are augmented during training with random gaussian noise within four distinct frequency ranges, each emulating real-world noise sources. Both models perform comparably on standard ECGs with an AUROC of 0.90. The noise-adapted model performs significantly better on the same test set augmented with four distinct real-world noise recordings at multiple signal-to-noise ratios (SNRs), including noise isolated from a portable device ECG. The standard and noise-adapted models have an AUROC of 0.72 and 0.87, respectively, when evaluated on ECGs augmented with portable ECG device noise at an SNR of 0.5. This approach represents a novel strategy for the development of wearable-adapted tools from clinical ECG repositories.

Abstract The benefits of large-scale genetic studies for healthcare of the populations studied are well documented, but these genetic studies have traditionally ignored people from some parts of the world, such as South Asia. Here we describe whole genome sequence (WGS) data from 4806 individuals recruited from the healthcare delivery systems of Pakistan, India and Bangladesh, combined with WGS from 927 individuals from isolated South Asian populations. We characterize population structure in South Asia and describe a genotyping array (SARGAM) and imputation reference panel that are optimized for South Asian genomes. We find evidence for high rates of reproductive isolation, endogamy and consanguinity that vary across the subcontinent and that lead to levels of rare homozygotes that reach 100 times that seen in outbred populations. Founder effects increase the power to associate functional variants with disease processes and make South Asia a uniquely powerful place for population-scale genetic studies.

Abstract Lipid nanoparticles have demonstrated utility in hepatic delivery of a range of therapeutic modalities and typically deliver their cargo via low-density lipoprotein receptor-mediated endocytosis. For patients lacking sufficient low-density lipoprotein receptor activity, such as those with homozygous familial hypercholesterolemia, an alternate strategy is needed. Here we show the use of structure-guided rational design in a series of mouse and non-human primate studies to optimize a GalNAc-Lipid nanoparticle that allows for low-density lipoprotein receptor independent delivery. In low-density lipoprotein receptor-deficient non-human primates administered a CRISPR base editing therapy targeting the ANGPTL3 gene, the introduction of an optimized GalNAc-based asialoglycoprotein receptor ligand to the nanoparticle surface increased liver editing from 5% to 61% with minimal editing in nontargeted tissues. Similar editing was noted in wild-type monkeys, with durable blood ANGPTL3 protein reduction up to 89% six months post dosing. These results suggest that GalNAc-Lipid nanoparticles may effectively deliver to both patients with intact low-density lipoprotein receptor activity as well as those afflicted by homozygous familial hypercholesterolemia.

Different location of adipose tissue may have different consequences to cardiometabolic risk. Here the authors report that deep learning enabled accurate prediction of specific adipose tissue volumes, and that after adjustment for BMI, visceral adiposity was associated with increased risk of cardiometabolic disease, while gluteofemoral adiposity was associated with reduced risk.

Abstract Prediction models are commonly used to estimate risk for cardiovascular diseases, to inform diagnosis and management. However, performance may vary substantially across relevant subgroups of the population. Here we investigated heterogeneity of accuracy and fairness metrics across a variety of subgroups for risk prediction of two common diseases: atrial fibrillation (AF) and atherosclerotic cardiovascular disease (ASCVD). We calculated the Cohorts for Heart and Aging in Genomic Epidemiology Atrial Fibrillation (CHARGE-AF) score for AF and the Pooled Cohort Equations (PCE) score for ASCVD in three large datasets: Explorys Life Sciences Dataset (Explorys, n = 21,809,334), Mass General Brigham (MGB, n = 520,868), and the UK Biobank (UKBB, n = 502,521). Our results demonstrate important performance heterogeneity across subpopulations defined by age, sex, and presence of preexisting disease, with fairly consistent patterns across both scores. For example, using CHARGE-AF, discrimination declined with increasing age, with a concordance index of 0.72 [95% CI 0.72–0.73] for the youngest (45–54 years) subgroup to 0.57 [0.56–0.58] for the oldest (85–90 years) subgroup in Explorys. Even though sex is not included in CHARGE-AF, the statistical parity difference (i.e., likelihood of being classified as high risk) was considerable between males and females within the 65–74 years subgroup with a value of − 0.33 [95% CI − 0.33 to − 0.33]. We also observed weak discrimination (i.e.,

Abstract Inter-individual variation in fat distribution is increasingly recognized as clinically important but is not routinely assessed in clinical practice, in part because medical imaging has not been practical to deploy at scale for this task. Here, we report a deep learning model trained on an individual’s body shape outline—or “silhouette” —that enables accurate estimation of specific fat depots of interest, including visceral (VAT), abdominal subcutaneous (ASAT), and gluteofemoral (GFAT) adipose tissue volumes, and VAT/ASAT ratio. Two-dimensional coronal and sagittal silhouettes are constructed from whole-body magnetic resonance images in 40,032 participants of the UK Biobank and used as inputs for a convolutional neural network to predict each of these quantities. Mean age of the study participants is 65 years and 51% are female. A cross-validated deep learning model trained on silhouettes enables accurate estimation of VAT, ASAT, and GFAT volumes (R 2: 0.88, 0.93, and 0.93, respectively), outperforming a comparator model combining anthropometric and bioimpedance measures (ΔR 2 = 0.05–0.13). Next, we study VAT/ASAT ratio, a nearly body-mass index (BMI)—and waist circumference-independent marker of metabolically unhealthy fat distribution. While the comparator model poorly predicts VAT/ASAT ratio (R 2: 0.17–0.26), a silhouette-based model enables significant improvement (R 2: 0.50–0.55). Increased silhouette-predicted VAT/ASAT ratio is associated with increased risk of prevalent and incident type 2 diabetes and coronary artery disease independent of BMI and waist circumference. These results demonstrate that body silhouette images can estimate important measures of fat distribution, laying the scientific foundation for scalable population-based assessment.

The inherited basis of body fat distribution is not fully understood. Here, the authors use genetic data and MRI-derived measures of local adiposity to highlight fat depot-specific genetic architecture with implications for cardiometabolic health.

Polygenic scores can identify individuals with high disease risk based on inborn DNA variation. We explore their potential to enrich clinical trials by identifying individuals based on higher risk of disease (‘prognostic enrichment’), or increased probability of benefit (‘predictive enrichment’).

Abstract Diagnosis codes are used to study SARS-CoV2 infections and COVID-19 hospitalizations in administrative and electronic health record (EHR) data. Using EHR data (April 2020–March 2021) at the Yale-New Haven Health System and the three hospital systems of the Mayo Clinic, computable phenotype definitions based on ICD-10 diagnosis of COVID-19 (U07.1) were evaluated against positive SARS-CoV-2 PCR or antigen tests. We included 69,423 patients at Yale and 75,748 at Mayo Clinic with either a diagnosis code or a positive SARS-CoV-2 test. The precision and recall of a COVID-19 diagnosis for a positive test were 68.8% and 83.3%, respectively, at Yale, with higher precision (95%) and lower recall (63.5%) at Mayo Clinic, varying between 59.2% in Rochester to 97.3% in Arizona. For hospitalizations with a principal COVID-19 diagnosis, 94.8% at Yale and 80.5% at Mayo Clinic had an associated positive laboratory test, with secondary diagnosis of COVID-19 identifying additional patients. These patients had a twofold higher inhospital mortality than based on principal diagnosis. Standardization of coding practices is needed before the use of diagnosis codes in clinical research and epidemiological surveillance of COVID-19.

The application of artificial intelligence for automated diagnosis of electrocardiograms can improve care in remote settings but is limited by the reliance on infrequently available signal-based data. Here, the authors report the development of a multi-label automated diagnosis model for electrocardiographic images.

Abstract There is little data describing trends in the use of hydroxychloroquine for COVID-19 following publication of randomized trials that failed to demonstrate a benefit of this therapy. We identified 13,957 patients admitted for active COVID-19 at 85 U.S. hospitals participating in a national registry between March 1 and August 31, 2020. The overall proportion of patients receiving hydroxychloroquine peaked at 55.2% in March and April and decreased to 4.8% in May and June and 0.8% in July and August. At the hospital-level, median use was 59.4% in March and April (IQR 48.5–71.5%, range 0–100%) and decreased to 0.3% (IQR 0–5.4%, range 0–100%) by May and June and 0% (IQR 0–1.3%, range 0–36.4%) by July and August. The rate and hospital-level uniformity in deimplementation of this ineffective therapy for COVID-19 reflects a rapid response to evolving clinical information and further study may offer strategies to inform deimplementation of ineffective clinical care.

Low dose computed tomography (LDCT) for lung cancer screening offers an opportunity for simultaneous CVD risk estimation in at-risk patients. Here, the authors develop a deep learning model to perform this task, showing human-level performance.

Genetic variation predisposes to disease via monogenic and polygenic risk variants. Here, the authors assess the interplay between these types of variation on disease penetrance in 80,928 individuals. In carriers of monogenic variants, they show that disease risk is a gradient influenced by polygenic background.

Structural changes to the left ventricle are characteristic of dilated cardiomyopathy (DCM), a disease for which many rare genetic variants are known. Here, Pirruccello et al. report GWAS of seven cardiac MRI measurements in the left ventricle and describe shared loci and polygenic association with DCM.

Abstract We hypothesized that clinical process improvement strategies can reduce frequency of motion artifacts and expiratory phase scanning in chest CT. We reviewed 826 chest CT to establish the baseline frequency. Per clinical process improvement guidelines, we brainstormed corrective measures and priority-pay-off matrix. The first intervention involved education of CT technologists, following which 795 chest CT were reviewed. For the second intervention, instructional videos on optimal breath-hold were shown to 245 adult patients just before their chest CT. Presence of motion artifacts and expiratory phase scanning was assessed. We also reviewed 311 chest CT scans belonging to a control group of patients who did not see the instructional videos. Pareto and percentage run charts were created for baseline and post-intervention data. Baseline incidence of motion artifacts and expiratory phase scanning in chest CT was 35% (292/826). There was no change in the corresponding incidence following the first intervention (36%; 283/795). Respiratory motion and expiratory phase chest CT with the second intervention decreased (8%, 20/245 patients). Instructional videos for patients (and not education and training of CT technologists) reduce the frequency of motion artifacts and expiratory phase scanning in chest CT.

Common genetic variants associated with plasma lipids have been extensively studied for a better understanding of common diseases. Here, the authors use whole-genome sequencing of 16,324 individuals to analyze rare variant associations and to determine their monogenic and polygenic contribution to lipid traits.

Genetic variation in ANGPTL4 is associated with lipid traits. Here, the authors find that predicted loss-of-function variants in ANGPTL4 are associated with glucose homeostasis and reduced risk of type 2 diabetes and that Angptl4 −/− mice on a high-fat diet show improved insulin sensitivity.

Examination of predicted loss-of-function (pLOF) genetic variants allows direct identification of genes with therapeutic potential. Here, Emdin et al. perform association analysis for 3759 pLOF variants with 24 traits and highlight protective variants against cardiometabolic and immune phenotypes.