Deep Learning Algorithms to Detect Murmurs Associated With Structural Heart Disease.

Background The success of cardiac auscultation varies widely among medical professionals, which can lead to missed treatments for structural heart disease. Applying machine learning to cardiac auscultation could address this problem, but despite recent interest, few algorithms have been brought to clinical practice. We evaluated a novel suite of Food and Drug Administration-cleared algorithms trained via deep learning on >15 000 heart sound recordings. Methods and Results We validated the algorithms on a data set of 2375 recordings from 615 unique subjects. This data set was collected in real clinical environments using commercially available digital stethoscopes, annotated by board-certified cardiologists, and paired with echocardiograms as the gold standard. To model the algorithm in clinical practice, we compared its performance against 10 clinicians on a subset of the validation database. Our algorithm reliably detected structural murmurs with a sensitivity of 85.6% and specificity of 84.4%. When limiting the analysis to clearly audible murmurs in adults, performance improved to a sensitivity of 97.9% and specificity of 90.6%. The algorithm also reported timing within the cardiac cycle, differentiating between systolic and diastolic murmurs. Despite optimizing acoustics for the clinicians, the algorithm substantially outperformed the clinicians (average clinician accuracy, 77.9%; algorithm accuracy, 84.7%.) Conclusions The algorithms accurately identified murmurs associated with structural heart disease. Our results illustrate a marked contrast between the consistency of the algorithm and the substantial interobserver variability of clinicians. Our results suggest that adopting machine learning algorithms into clinical practice could improve the detection of structural heart disease to facilitate patient care.

Social Determinants of Emergency Department Visits in Mild Compared to Moderate and Severe Asthma.

BACKGROUND: Social determinants of health (SDH) influence emergency department (ED) use among children with asthma. We aimed to examine if SDH were more strongly associated with ED use among children with moderate/severe compared to mild asthma. METHODS: This study utilized the 2016-2019 data from the National Survey of Children's Health. Children with asthma ages 0-17 years (N = 9937) were included in the analysis. Asthma severity and all-cause ED use in the past year were reported by caregivers. The association between patient factors and ED visits was evaluated using ordinal logistic regression. RESULTS: Based on the study sample, 29% of children with asthma had moderate/severe asthma. In the mild group, 30% visited the ED at least once in the past 12 months, compared to 49% in the moderate/severe group. SDH associated with ED visits included race/ethnicity, insurance coverage, and parental educational attainment, but the strength of these associations did not vary according to asthma severity. CONCLUSIONS: In a nationally representative data set, SDH were equally predictive of ED use regardless of children's asthma severity. Interventions to reduce ED use among children with asthma should be considered for children with any severity of asthma, especially children in socially disadvantaged groups at higher risk of ED utilization.

Proactive Lactation Care is Associated With Improved Outcomes in a Referral NICU.

BACKGROUND: Mother's milk improves outcomes. Referral neonatal intensive care units face unique lactation challenges with maternal-infant separation and maternal pump dependency. Little is known about lactation resource allocation in this high-risk population. RESEARCH AIMS: To determine differences in human milk outcomes, (1) the proportion of infants fed exclusive or any mother's milk and (2) recorded number and volume of pumped mothers' milk bottles, between two models of lactation care in a referral neonatal intensive care unit. METHODS: This retrospective, longitudinal, two-group comparison study utilized medical record individual feeding data for infants admitted at ≤ Day 7 of age and milk room storage records from reactive and proactive care model time periods (April, 2017-March, 2018; May, 2018-April, 2019). The reactive care model (n = 509 infants, 58% male, median birth weight and gestational age of 37 weeks,) involved International Board Certified Lactation Consultant referral for identified lactation problems; whereas, the proactive model (n = 472 infants, 56% male, median birth weight and gestational age 37 weeks) increased International Board Certified Lactation Consultant staffing, who then saw all admissions. Comparisons were performed using chi square, Mann Whitney, and t-tests. RESULTS: A proactive lactation approach was associated with an increase in the receipt of any mother's milk from 74.3% to 80.2% (p = .03) among participants in the proactive model group. Additionally, their milk room mean monthly bottle storage increased from 5153 (SD 788) to 6620 (SD 1314) bottles (p < .01). CONCLUSIONS: In this retrospective study at a tertiary referral neonatal intensive care unit, significant improvement inhuman milk outcomes suggests that increased resources for proactive lactation care may improve mother's milk provision for a high-risk population.

Deep Learning Algorithm for Automated Cardiac Murmur Detection via a Digital Stethoscope Platform.

Background Clinicians vary markedly in their ability to detect murmurs during cardiac auscultation and identify the underlying pathological features. Deep learning approaches have shown promise in medicine by transforming collected data into clinically significant information. The objective of this research is to assess the performance of a deep learning algorithm to detect murmurs and clinically significant valvular heart disease using recordings from a commercial digital stethoscope platform. Methods and Results Using >34 hours of previously acquired and annotated heart sound recordings, we trained a deep neural network to detect murmurs. To test the algorithm, we enrolled 962 patients in a clinical study and collected recordings at the 4 primary auscultation locations. Ground truth was established using patient echocardiograms and annotations by 3 expert cardiologists. Algorithm performance for detecting murmurs has sensitivity and specificity of 76.3% and 91.4%, respectively. By omitting softer murmurs, those with grade 1 intensity, sensitivity increased to 90.0%. Application of the algorithm at the appropriate anatomic auscultation location detected moderate-to-severe or greater aortic stenosis, with sensitivity of 93.2% and specificity of 86.0%, and moderate-to-severe or greater mitral regurgitation, with sensitivity of 66.2% and specificity of 94.6%. Conclusions The deep learning algorithm's ability to detect murmurs and clinically significant aortic stenosis and mitral regurgitation is comparable to expert cardiologists based on the annotated subset of our database. The findings suggest that such algorithms would have utility as front-line clinical support tools to aid clinicians in screening for cardiac murmurs caused by valvular heart disease. Registration URL: https://clinicaltrials.gov; Unique Identifier: NCT03458806.

Swine acute diarrhea syndrome coronavirus replication in primary human cells reveals potential susceptibility to infection.

Zoonotic coronaviruses represent an ongoing threat, yet the myriads of circulating animal viruses complicate the identification of higher-risk isolates that threaten human health. Swine acute diarrhea syndrome coronavirus (SADS-CoV) is a newly discovered, highly pathogenic virus that likely evolved from closely related HKU2 bat coronaviruses, circulating in Rhinolophus spp. bats in China and elsewhere. As coronaviruses cause severe economic losses in the pork industry and swine are key intermediate hosts of human disease outbreaks, we synthetically resurrected a recombinant virus (rSADS-CoV) as well as a derivative encoding tomato red fluorescent protein (tRFP) in place of ORF3. rSADS-CoV replicated efficiently in a variety of continuous animal and primate cell lines, including human liver and rectal carcinoma cell lines. Of concern, rSADS-CoV also replicated efficiently in several different primary human lung cell types, as well as primary human intestinal cells. rSADS-CoV did not use human coronavirus ACE-2, DPP4, or CD13 receptors for docking and entry. Contemporary human donor sera neutralized the group I human coronavirus NL63, but not rSADS-CoV, suggesting limited human group I coronavirus cross protective herd immunity. Importantly, remdesivir, a broad-spectrum nucleoside analog that is effective against other group 1 and 2 coronaviruses, efficiently blocked rSADS-CoV replication in vitro. rSADS-CoV demonstrated little, if any, replicative capacity in either immune-competent or immunodeficient mice, indicating a critical need for improved animal models. Efficient growth in primary human lung and intestinal cells implicate SADS-CoV as a potential higher-risk emerging coronavirus pathogen that could negatively impact the global economy and human health.

Sumoylation regulates interaction of FOG1 with C-terminal-binding protein (CTBP).

Erythropoietic and megakaryocytic programs are specified from multipotential progenitors by the transcription factor GATA1. FOG1, a GATA1-interaction partner, is critical for GATA1 function in several contexts by bringing multiple complexes into association with GATA1 to facilitate activation or repression of target genes. To further elucidate regulation of these associations by cellular and extracellular cues, we examined FOG1 for post-translational modifications. We found that FOG1 is SUMOylated and phosphorylated in erythroid cells in a differentiation-dependent manner. Removal of the SUMOylation sites in FOG1 does not impair nuclear localization, protein stability, or chromatin occupancy. However, SUMOylation of FOG1 modulates interactions with C-terminal binding protein family members, specifically promoting CTBP1 binding. Phosphorylation of FOG1 modulates SUMOylation and, therefore, indirectly regulates the CTBP interaction. Post-translational modification of FOG1 may contribute to control of co-occupancy by CTBP family members, the NuRD complex, and GATA1 at differentially regulated genes.

Cogan's syndrome complicating pregnancy.

Cogan's syndrome is a rare autoimmune disorder characterized by systemic vasculitis as well as eye and ear inflammation. A 33-year-old G1 P0 with Cogan's syndrome experienced an uneventful pregnancy while being treated with hydroxychloroquine and prednisone. In the absence of systemic vasculitis, Cogan's syndrome may be associated with favorable pregnancy outcome.

Transfusion of platelet components prepared with photochemical pathogen inactivation treatment during a Chikungunya virus epidemic in Ile de La Réunion.

BACKGROUND: During the Chikungunya virus (CHIKV) epidemic on Ile de La Réunion, France, more than 30% of 750,000 inhabitants were infected. Local blood donation was suspended to prevent transfusion-transmitted infection (TT-CHIKV). To sustain the availability of platelet (PLT) components, the Etablissement Français du Sang implemented universal pathogen inactivation (INTERCEPT, Cerus Europe BV) of PLT components (CPAs). The study assessed the safety of PLT components treated with pathogen inactivation transfused in routine clinical practice. STUDY DESIGN AND METHODS: This was a retrospective observational study using patient medical records and the AFSSAPS hemovigilance database (eFIT) to identify TT-CHIKV and adverse events (AEs) classified as acute transfusion reactions (ATRs) to PLT components prepared with pathogen inactivation. RESULTS: During 1 year, 1950 INTERCEPT-CPAs were transfused to 335 adult, 51 pediatric, and 41 infant patients. Nineteen AEs were observed in 15 patients and 10 were classified as ATRs. Eight ATRs occurred in 6 pediatric hematology-oncology patients. No ATRs were observed in infants. The most frequently reported signs and symptoms were Grade 1 urticaria, itching, chills, fever, and anxiety. No cases of transfusion-related acute lung injury, TT-sepsis, or TT-CHIKV were detected. CONCLUSIONS: INTERCEPT-CPAs were well tolerated in a broad range of patients, including infants. ATR incidence was low and when present ATRs were of mild severity.

Generation of iPS cells using defined factors linked via the self-cleaving 2A sequences in a single open reading frame.

Generation of induced pluripotent stem (iPS) cells from somatic cells has been achieved successfully by simultaneous viral transduction of defined reprogramming transcription factors (TFs). However, the process requires multiple viral vectors for gene delivery. As a result, generated iPS cells harbor numerous viral integration sites in their genomes. This can increase the probability of gene mutagenesis and genomic instability, and present significant barriers to both research and clinical application studies of iPS cells. In this paper, we present a simple lentivirus reprogramming system in which defined factors are fused in-frame into a single open reading frame (ORF) via self-cleaving 2A sequences. A GFP marker is placed downstream of the transgene to enable tracking of transgene expression. We demonstrate that this polycistronic expression system efficiently generates iPS cells. The generated iPS cells have normal karyotypes and are similar to mouse embryonic stem cells in morphology and gene expression. Moreover, they can differentiate into cell types of the three embryonic germ layers in both in vitro and in vivo assays. Remarkably, most of these iPS cells only harbor a single copy of viral vector. This system provides a valuable tool for generation of iPS cells, and our data suggest that the balance of expression of transduced reprogramming TFs in each cell is essential for the reprogramming process. More importantly, when delivered by non-integrating gene-delivery systems, this re-engineered single ORF will facilitate efficient generation of human iPS cells free of genetic modifications.