Variational Disentanglement for Rare Event Modeling.

Combining the increasing availability and abundance of healthcare data and the current advances in machine learning methods have created renewed opportunities to improve clinical decision support systems. However, in healthcare risk prediction applications, the proportion of cases with the condition (label) of interest is often very low relative to the available sample size. Though very prevalent in healthcare, such imbalanced classification settings are also common and challenging in many other scenarios. So motivated, we propose a variational disentanglement approach to semi-parametrically learn from rare events in heavily imbalanced classification problems. Specifically, we leverage the imposed extreme-distribution behavior on a latent space to extract information from low-prevalence events, and develop a robust prediction arm that joins the merits of the generalized additive model and isotonic neural nets. Results on synthetic studies and diverse real-world datasets, including mortality prediction on a COVID-19 cohort, demonstrate that the proposed approach outperforms existing alternatives.

Blood RNA alternative splicing events as diagnostic biomarkers for infectious disease

Summary Assays detecting blood transcriptome changes are studied for infectious disease diagnosis. Blood-based RNA alternative splicing (AS) events, which have not been well characterized in pathogen infection, have potential normalization and assay platform stability advantages over gene expression for diagnosis. Here, we present a computational framework for developing AS diagnostic biomarkers. Leveraging a large prospective cohort of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and whole-blood RNA sequencing (RNA-seq) data, we identify a major functional AS program switch upon viral infection. Using an independent cohort, we demonstrate the improved accuracy of AS biomarkers for SARS-CoV-2 diagnosis compared with six reported transcriptome signatures. We then optimize a subset of AS-based biomarkers to develop microfluidic PCR diagnostic assays. This assay achieves nearly perfect test accuracy (61/62 = 98.4%) using a naive principal component classifier, significantly more accurate than a gene expression PCR assay in the same cohort. Therefore, our RNA splicing computational framework enables a promising avenue for host-response diagnosis of infection. Graphical abstract Highlights • We present a computational framework for alternative splicing (AS) diagnostic markers• Our AS biomarkers outperform gene-expression biomarkers in COVID-19 detection• Microfluidic PCR diagnostic assay of AS biomarkers achieves greater than 98% accuracy• We interpret the biological importance of identified AS biomarkers Motivation Host-based response assays (HRAs) can often diagnose infectious disease earlier and more precisely than pathogen-based tests. However, the role of RNA alternative splicing (AS) in HRAs remains unexplored, as existing HRAs are restricted to gene expression signatures. We report a computational framework for the identification, optimization, and evaluation of blood AS-based diagnostic assay development for infectious disease. Using SARS-CoV-2 infection as a case study, we demonstrate the improved accuracy of AS biomarkers for COVID-19 diagnosis when compared against six reported transcriptome signatures and when implemented as a microfluidic PCR diagnostic assay. Host-based response assays can diagnose infectious disease earlier and more precisely than pathogen-based tests. However, the role of RNA alternative splicing (AS) remains unexplored. Zhang et al. present a computational framework for AS diagnostic biomarkers. Using SARS-CoV-2 as a case study, they demonstrate the improved accuracy of AS biomarkers for COVID-19 diagnosis.

Blood RNA alternative splicing events as diagnostic biomarkers for infectious disease.

Assays detecting blood transcriptome changes are studied for infectious disease diagnosis. Blood-based RNA alternative splicing (AS) events, which have not been well characterized in pathogen infection, have potential normalization and assay platform stability advantages over gene expression for diagnosis. Here, we present a computational framework for developing AS diagnostic biomarkers. Leveraging a large prospective cohort of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and whole-blood RNA sequencing (RNA-seq) data, we identify a major functional AS program switch upon viral infection. Using an independent cohort, we demonstrate the improved accuracy of AS biomarkers for SARS-CoV-2 diagnosis compared with six reported transcriptome signatures. We then optimize a subset of AS-based biomarkers to develop microfluidic PCR diagnostic assays. This assay achieves nearly perfect test accuracy (61/62 = 98.4%) using a naive principal component classifier, significantly more accurate than a gene expression PCR assay in the same cohort. Therefore, our RNA splicing computational framework enables a promising avenue for host-response diagnosis of infection.

Multi-objective optimization identifies a specific and interpretable COVID-19 host response signature.

The identification of a COVID-19 host response signature in blood can increase the understanding of SARS-CoV-2 pathogenesis and improve diagnostic tools. Applying a multi-objective optimization framework to both massive public and new multi-omics data, we identified a COVID-19 signature regulated at both transcriptional and epigenetic levels. We validated the signature's robustness in multiple independent COVID-19 cohorts. Using public data from 8,630 subjects and 53 conditions, we demonstrated no cross-reactivity with other viral and bacterial infections, COVID-19 comorbidities, or confounders. In contrast, previously reported COVID-19 signatures were associated with significant cross-reactivity. The signature's interpretation, based on cell-type deconvolution and single-cell data analysis, revealed prominent yet complementary roles for plasmablasts and memory T cells. Although the signal from plasmablasts mediated COVID-19 detection, the signal from memory T cells controlled against cross-reactivity with other viral infections. This framework identified a robust, interpretable COVID-19 signature and is broadly applicable in other disease contexts. A record of this paper's transparent peer review process is included in the supplemental information.

Systematic comparison of published host gene expression signatures for bacterial/viral discrimination.

BACKGROUND: Measuring host gene expression is a promising diagnostic strategy to discriminate bacterial and viral infections. Multiple signatures of varying size, complexity, and target populations have been described. However, there is little information to indicate how the performance of various published signatures compare to one another. METHODS: This systematic comparison of host gene expression signatures evaluated the performance of 28 signatures, validating them in 4589 subjects from 51 publicly available datasets. Thirteen COVID-specific datasets with 1416 subjects were included in a separate analysis. Individual signature performance was evaluated using the area under the receiving operating characteristic curve (AUC) value. Overall signature performance was evaluated using median AUCs and accuracies. RESULTS: Signature performance varied widely, with median AUCs ranging from 0.55 to 0.96 for bacterial classification and 0.69-0.97 for viral classification. Signature size varied (1-398 genes), with smaller signatures generally performing more poorly (P < 0.04). Viral infection was easier to diagnose than bacterial infection (84% vs. 79% overall accuracy, respectively; P < .001). Host gene expression classifiers performed more poorly in some pediatric populations (3 months-1 year and 2-11 years) compared to the adult population for both bacterial infection (73% and 70% vs. 82%, respectively; P < .001) and viral infection (80% and 79% vs. 88%, respectively; P < .001). We did not observe classification differences based on illness severity as defined by ICU admission for bacterial or viral infections. The median AUC across all signatures for COVID-19 classification was 0.80 compared to 0.83 for viral classification in the same datasets. CONCLUSIONS: In this systematic comparison of 28 host gene expression signatures, we observed differences based on a signature's size and characteristics of the validation population, including age and infection type. However, populations used for signature discovery did not impact performance, underscoring the redundancy among many of these signatures. Furthermore, differential performance in specific populations may only be observable through this type of large-scale validation.

Lesion identification and malignancy prediction from clinical dermatological images.

We consider machine-learning-based lesion identification and malignancy prediction from clinical dermatological images, which can be indistinctly acquired via smartphone or dermoscopy capture. Additionally, we do not assume that images contain single lesions, thus the framework supports both focal or wide-field images. Specifically, we propose a two-stage approach in which we first identify all lesions present in the image regardless of sub-type or likelihood of malignancy, then it estimates their likelihood of malignancy, and through aggregation, it also generates an image-level likelihood of malignancy that can be used for high-level screening processes. Further, we consider augmenting the proposed approach with clinical covariates (from electronic health records) and publicly available data (the ISIC dataset). Comprehensive experiments validated on an independent test dataset demonstrate that (1) the proposed approach outperforms alternative model architectures; (2) the model based on images outperforms a pure clinical model by a large margin, and the combination of images and clinical data does not significantly improves over the image-only model; and (3) the proposed framework offers comparable performance in terms of malignancy classification relative to three board certified dermatologists with different levels of experience.

Telehealth Made EASY: Understanding Provider Perceptions of Telehealth Appropriateness in Outpatient Rheumatology Encounters.

OBJECTIVE: The purpose of this study was to evaluate a novel scoring system, the Encounter Appropriateness Score for You (EASY), to assess provider perceptions of telehealth appropriateness in rheumatology encounters. METHODS: The EASY scoring system prompts providers to rate their own encounters as follows: in-person or telehealth acceptable, EASY = 1; in-person preferred, EASY = 2; or telehealth preferred, EASY = 3. Assessment of the EASY scoring system occurred at a single academic institution from January 1, 2021, to August 31, 2021. Data were collected in three rounds: 1) initial survey (31 providers) assessing EASY responsiveness to five hypothetical scenarios, 2) follow-up survey (34 providers) exploring EASY responsiveness to 11 scenario modifications, and 3) assessment of EASYs documented in clinic care. RESULTS: The initial and follow-up surveys demonstrated responsiveness of EASYs to different clinical and nonclinical factors. For instance, less than 20% of providers accepted telehealth when starting a biologic for active rheumatoid arthritis, although more than 35% accepted telehealth in the same scenario if the patient lived far away or was well known to the provider. Regarding EASY documentation, 27 providers provided EASYs for 12,381 encounters. According to these scores, telehealth was acceptable or preferred for 29.7% of all encounters, including 21.4% of in-person encounters. Conversely, 24.4% of telehealth encounters were scored as in-person preferred. CONCLUSION: EASY is simple, understandable, and responsive to changes in the clinical scenario. We have successfully accumulated 12,381 EASYs that can be studied in future work to better understand telehealth utility and optimize telehealth triage.

Differential chromatin accessibility in peripheral blood mononuclear cells underlies COVID-19 disease severity prior to seroconversion.

SARS-CoV-2 infection triggers profound and variable immune responses in human hosts. Chromatin remodeling has been observed in individuals severely ill or convalescing with COVID-19, but chromatin remodeling early in disease prior to anti-spike protein IgG seroconversion has not been defined. We performed the Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) and RNA-seq on peripheral blood mononuclear cells (PBMCs) from outpatients with mild or moderate symptom severity at different stages of clinical illness. Early in the disease course prior to IgG seroconversion, modifications in chromatin accessibility associated with mild or moderate symptoms were already robust and included severity-associated changes in accessibility of genes in interleukin signaling, regulation of cell differentiation and cell morphology. Furthermore, single-cell analyses revealed evolution of the chromatin accessibility landscape and transcription factor motif accessibility for individual PBMC cell types over time. The most extensive remodeling occurred in CD14+ monocytes, where sub-populations with distinct chromatin accessibility profiles were observed prior to seroconversion. Mild symptom severity was marked by upregulation of classical antiviral pathways, including those regulating IRF1 and IRF7, whereas in moderate disease, these classical antiviral signals diminished, suggesting dysregulated and less effective responses. Together, these observations offer novel insight into the epigenome of early mild SARS-CoV-2 infection and suggest that detection of chromatin remodeling in early disease may offer promise for a new class of diagnostic tools for COVID-19.

Prospective Validation of a Rapid Host Gene Expression Test to Discriminate Bacterial From Viral Respiratory Infection.

Importance: Bacterial and viral causes of acute respiratory illness (ARI) are difficult to clinically distinguish, resulting in the inappropriate use of antibacterial therapy. The use of a host gene expression-based test that is able to discriminate bacterial from viral infection in less than 1 hour may improve care and antimicrobial stewardship. Objective: To validate the host response bacterial/viral (HR-B/V) test and assess its ability to accurately differentiate bacterial from viral infection among patients with ARI. Design, Setting, and Participants: This prospective multicenter diagnostic study enrolled 755 children and adults with febrile ARI of 7 or fewer days' duration from 10 US emergency departments. Participants were enrolled from October 3, 2014, to September 1, 2019, followed by additional enrollment of patients with COVID-19 from March 20 to December 3, 2020. Clinical adjudication of enrolled participants identified 616 individuals as having bacterial or viral infection. The primary analysis cohort included 334 participants with high-confidence reference adjudications (based on adjudicator concordance and the presence of an identified pathogen confirmed by microbiological testing). A secondary analysis of the entire cohort of 616 participants included cases with low-confidence reference adjudications (based on adjudicator discordance or the absence of an identified pathogen in microbiological testing). Thirty-three participants with COVID-19 were included post hoc. Interventions: The HR-B/V test quantified the expression of 45 host messenger RNAs in approximately 45 minutes to derive a probability of bacterial infection. Main Outcomes and Measures: Performance characteristics for the HR-B/V test compared with clinical adjudication were reported as either bacterial or viral infection or categorized into 4 likelihood groups (viral very likely [probability score <0.19], viral likely [probability score of 0.19-0.40], bacterial likely [probability score of 0.41-0.73], and bacterial very likely [probability score >0.73]) and compared with procalcitonin measurement. Results: Among 755 enrolled participants, the median age was 26 years (IQR, 16-52 years); 360 participants (47.7%) were female, and 395 (52.3%) were male. A total of 13 participants (1.7%) were American Indian, 13 (1.7%) were Asian, 368 (48.7%) were Black, 131 (17.4%) were Hispanic, 3 (0.4%) were Native Hawaiian or Pacific Islander, 297 (39.3%) were White, and 60 (7.9%) were of unspecified race and/or ethnicity. In the primary analysis involving 334 participants, the HR-B/V test had sensitivity of 89.8% (95% CI, 77.8%-96.2%), specificity of 82.1% (95% CI, 77.4%-86.6%), and a negative predictive value (NPV) of 97.9% (95% CI, 95.3%-99.1%) for bacterial infection. In comparison, the sensitivity of procalcitonin measurement was 28.6% (95% CI, 16.2%-40.9%; P < .001), the specificity was 87.0% (95% CI, 82.7%-90.7%; P = .006), and the NPV was 87.6% (95% CI, 85.5%-89.5%; P < .001). When stratified into likelihood groups, the HR-B/V test had an NPV of 98.9% (95% CI, 96.1%-100%) for bacterial infection in the viral very likely group and a positive predictive value of 63.4% (95% CI, 47.2%-77.9%) for bacterial infection in the bacterial very likely group. The HR-B/V test correctly identified 30 of 33 participants (90.9%) with acute COVID-19 as having a viral infection. Conclusions and Relevance: In this study, the HR-B/V test accurately discriminated bacterial from viral infection among patients with febrile ARI and was superior to procalcitonin measurement. The findings suggest that an accurate point-of-need host response test with high NPV may offer an opportunity to improve antibiotic stewardship and patient outcomes.

Assessment of the Feasibility of Using Noninvasive Wearable Biometric Monitoring Sensors to Detect Influenza and the Common Cold Before Symptom Onset.

Importance: Currently, there are no presymptomatic screening methods to identify individuals infected with a respiratory virus to prevent disease spread and to predict their trajectory for resource allocation. Objective: To evaluate the feasibility of using noninvasive, wrist-worn wearable biometric monitoring sensors to detect presymptomatic viral infection after exposure and predict infection severity in patients exposed to H1N1 influenza or human rhinovirus. Design, Setting, and Participants: The cohort H1N1 viral challenge study was conducted during 2018; data were collected from September 11, 2017, to May 4, 2018. The cohort rhinovirus challenge study was conducted during 2015; data were collected from September 14 to 21, 2015. A total of 39 adult participants were recruited for the H1N1 challenge study, and 24 adult participants were recruited for the rhinovirus challenge study. Exclusion criteria for both challenges included chronic respiratory illness and high levels of serum antibodies. Participants in the H1N1 challenge study were isolated in a clinic for a minimum of 8 days after inoculation. The rhinovirus challenge took place on a college campus, and participants were not isolated. Exposures: Participants in the H1N1 challenge study were inoculated via intranasal drops of diluted influenza A/California/03/09 (H1N1) virus with a mean count of 106 using the median tissue culture infectious dose (TCID50) assay. Participants in the rhinovirus challenge study were inoculated via intranasal drops of diluted human rhinovirus strain type 16 with a count of 100 using the TCID50 assay. Main Outcomes and Measures: The primary outcome measures included cross-validated performance metrics of random forest models to screen for presymptomatic infection and predict infection severity, including accuracy, precision, sensitivity, specificity, F1 score, and area under the receiver operating characteristic curve (AUC). Results: A total of 31 participants with H1N1 (24 men [77.4%]; mean [SD] age, 34.7 [12.3] years) and 18 participants with rhinovirus (11 men [61.1%]; mean [SD] age, 21.7 [3.1] years) were included in the analysis after data preprocessing. Separate H1N1 and rhinovirus detection models, using only data on wearble devices as input, were able to distinguish between infection and noninfection with accuracies of up to 92% for H1N1 (90% precision, 90% sensitivity, 93% specificity, and 90% F1 score, 0.85 [95% CI, 0.70-1.00] AUC) and 88% for rhinovirus (100% precision, 78% sensitivity, 100% specificity, 88% F1 score, and 0.96 [95% CI, 0.85-1.00] AUC). The infection severity prediction model was able to distinguish between mild and moderate infection 24 hours prior to symptom onset with an accuracy of 90% for H1N1 (88% precision, 88% sensitivity, 92% specificity, 88% F1 score, and 0.88 [95% CI, 0.72-1.00] AUC) and 89% for rhinovirus (100% precision, 75% sensitivity, 100% specificity, 86% F1 score, and 0.95 [95% CI, 0.79-1.00] AUC). Conclusions and Relevance: This cohort study suggests that the use of a noninvasive, wrist-worn wearable device to predict an individual's response to viral exposure prior to symptoms is feasible. Harnessing this technology would support early interventions to limit presymptomatic spread of viral respiratory infections, which is timely in the era of COVID-19.

An atlas connecting shared genetic architecture of human diseases and molecular phenotypes provides insight into COVID-19 susceptibility.

BACKGROUND: While genome-wide associations studies (GWAS) have successfully elucidated the genetic architecture of complex human traits and diseases, understanding mechanisms that lead from genetic variation to pathophysiology remains an important challenge. Methods are needed to systematically bridge this crucial gap to facilitate experimental testing of hypotheses and translation to clinical utility. RESULTS: Here, we leveraged cross-phenotype associations to identify traits with shared genetic architecture, using linkage disequilibrium (LD) information to accurately capture shared SNPs by proxy, and calculate significance of enrichment. This shared genetic architecture was examined across differing biological scales through incorporating data from catalogs of clinical, cellular, and molecular GWAS. We have created an interactive web database (interactive Cross-Phenotype Analysis of GWAS database (iCPAGdb)) to facilitate exploration and allow rapid analysis of user-uploaded GWAS summary statistics. This database revealed well-known relationships among phenotypes, as well as the generation of novel hypotheses to explain the pathophysiology of common diseases. Application of iCPAGdb to a recent GWAS of severe COVID-19 demonstrated unexpected overlap of GWAS signals between COVID-19 and human diseases, including with idiopathic pulmonary fibrosis driven by the DPP9 locus. Transcriptomics from peripheral blood of COVID-19 patients demonstrated that DPP9 was induced in SARS-CoV-2 compared to healthy controls or those with bacterial infection. Further investigation of cross-phenotype SNPs associated with both severe COVID-19 and other human traits demonstrated colocalization of the GWAS signal at the ABO locus with plasma protein levels of a reported receptor of SARS-CoV-2, CD209 (DC-SIGN). This finding points to a possible mechanism whereby glycosylation of CD209 by ABO may regulate COVID-19 disease severity. CONCLUSIONS: Thus, connecting genetically related traits across phenotypic scales links human diseases to molecular and cellular measurements that can reveal mechanisms and lead to novel biomarkers and therapeutic approaches. The iCPAGdb web portal is accessible at http://cpag.oit.duke.edu and the software code at https://github.com/tbalmat/iCPAGdb .

Dysregulated transcriptional responses to SARS-CoV-2 in the periphery.

SARS-CoV-2 infection has been shown to trigger a wide spectrum of immune responses and clinical manifestations in human hosts. Here, we sought to elucidate novel aspects of the host response to SARS-CoV-2 infection through RNA sequencing of peripheral blood samples from 46 subjects with COVID-19 and directly comparing them to subjects with seasonal coronavirus, influenza, bacterial pneumonia, and healthy controls. Early SARS-CoV-2 infection triggers a powerful transcriptomic response in peripheral blood with conserved components that are heavily interferon-driven but also marked by indicators of early B-cell activation and antibody production. Interferon responses during SARS-CoV-2 infection demonstrate unique patterns of dysregulated expression compared to other infectious and healthy states. Heterogeneous activation of coagulation and fibrinolytic pathways are present in early COVID-19, as are IL1 and JAK/STAT signaling pathways, which persist into late disease. Classifiers based on differentially expressed genes accurately distinguished SARS-CoV-2 infection from other acute illnesses (auROC 0.95 [95% CI 0.92-0.98]). The transcriptome in peripheral blood reveals both diverse and conserved components of the immune response in COVID-19 and provides for potential biomarker-based approaches to diagnosis.

Impact of the COVID-19 pandemic on patterns of outpatient cardiovascular care.

BACKGROUND: The coronavirus disease 2019 (COVID-19) pandemic brought about abrupt changes in the way health care is delivered, and the impact of transitioning outpatient clinic visits to telehealth visits on processes of care and outcomes is unclear. METHODS: We evaluated ordering patterns during cardiovascular telehealth clinic visits in the Duke University Health System between March 15 and June 30, 2020 and 30-day outcomes compared with in-person visits in the same time frame in 2020 and in 2019. RESULTS: Within the Duke University Health System, there was a 33.1% decrease in the number of outpatient cardiovascular visits conducted in the first 15 weeks of the COVID-19 pandemic, compared with the same time period in 2019. As a proportion of total visits initially booked, 53% of visits were cancelled in 2020 compared to 35% in 2019. However, patients with cancelled visits had similar demographics and comorbidities in 2019 and 2020. Telehealth visits comprised 9.3% of total visits initially booked in 2020, with younger and healthier patients utilizing telehealth compared with those utilizing in-person visits. Compared with in-person visits in 2020, telehealth visits were associated with fewer new (31.6% for telehealth vs 44.6% for in person) or refill (12.9% vs 15.6%, respectively) medication prescriptions, electrocardiograms (4.3% vs 31.4%), laboratory orders (5.9% vs 21.8%), echocardiograms (7.3% vs 98%), and stress tests (4.4% vs 6.6%). When adjusted for age, race, and insurance status, those who had a telehealth visit or cancelled their visit were less likely to have an emergency department or hospital encounter within 30 days compared with those who had in-person visits (adjusted rate ratios (aRR) 0.76 [95% 0.65, 0.89] and aRR 0.71 [95% 0.65, 0.78], respectively). CONCLUSIONS: In response to the perceived risks of routine medical care affected by the COVID-19 pandemic, different phenotypes of patients chose different types of outpatient cardiology care. A better understanding of these differences could help define necessary and appropriate mode of care for cardiology patients.