Search

This study investigates the potential of using synthetic text to augment training data for Natural Language Processing (NLP) models, specifically within the context of peer support tools. We surveyed 22 participants-13 professional peer supporters and 9 AI-proficient individuals-tasked with distinguishing between AI-generated and human-written sentences. Using signal detection theory and confidence-based metrics, we evaluated the accuracy and confidence levels of both groups. The results show no significant differences in rater agreement between the two groups (p = 0.116), with overall classification accuracy falling below chance levels (mean accuracy = 43.10%, p < 0.001). Both groups exhibited a tendency to misclassify low-fidelity sentences as AI-generated, with peer supporters showing a significant bias (p = 0.007). Further analysis revealed a significant negative correlation between errors and confidence among AI-proficient raters (r = -0.429, p < 0.001), suggesting that as their confidence increased, their error rates decreased. Our findings support the feasibility of using synthetic text to mimic human communication, with important implications for improving the fidelity of peer support interventions through NLP model development.

Non-human primates remain the most useful and reliable pre-clinical model for many human diseases. Primate breath profiles have previously distinguished healthy animals from diseased, including non-human primates. Breath collection is relatively non-invasive, so this motivated us to define a healthy baseline breath profile that could be used in studies evaluating disease, therapies, and vaccines in non-human primates. A pilot study, which enrolled 30 healthy macaques, was conducted. Macaque breath molecules were sampled into a Tedlar bag, concentrated onto a thermal desorption tube, then desorbed and analyzed by comprehensive two-dimensional gas chromatography-time of flight mass spectrometry. These breath samples contained 2,017 features, of which 113 molecules were present in all breath samples. The core breathprint was dominated by aliphatic hydrocarbons, aromatic compounds, and carbonyl compounds. The data were internally validated with additional breath samples from a subset of 19 of these non-human primates. A critical core consisting of 23 highly abundant and invariant molecules was identified as a pragmatic breathprint set, useful for future validation studies in healthy primates., (© 2024. The Author(s).)

Confirmatory diagnosis of childhood tuberculosis (TB) remains a challenge mainly due to its dependence on sputum samples and the paucibacillary nature of the disease. Thus, only ~ 30% of suspected cases in children are diagnosed and the need for minimally invasive, non-sputum-based biomarkers remains unmet. Understanding host molecular changes by measuring blood-based transcriptomic markers has shown promise as a diagnostic tool for TB. However, the implication of sex contributing to disease heterogeneity and therefore diagnosis remains to be understood. Using publicly available gene expression data (GSE39939, GSE39940; n = 370), we report a sex-specific RNA biomarker signature that could improve the diagnosis of TB disease in children. We found four gene biomarker signatures for male (SLAMF8, GBP2, WARS, and FCGR1C) and female pediatric patients (GBP6, CELSR3, ALDH1A1, and GBP4) from Kenya, South Africa, and Malawi. Both signatures achieved a sensitivity of 85% and a specificity of 70%, which approaches the WHO-recommended target product profile for a triage test. Our gene signatures outperform most other gene signatures reported previously for childhood TB diagnosis., (© 2024. The Author(s).)

The availability of multiple publicly-available datasets studying the same phenomenon has the promise of accelerating scientific discovery. Meta-analysis can address issues of reproducibility and often increase power. The promise of meta-analysis is especially germane to rarer diseases like cystic fibrosis (CF), which affects roughly 100,000 people worldwide. A recent search of the National Institute of Health's Gene Expression Omnibus revealed 1.3 million data sets related to cancer compared to about 2,000 related to CF. These studies are highly diverse, involving different tissues, animal models, treatments, and clinical covariates. In our search for gene expression studies of primary human airway epithelial cells, we identified three studies with compatible methodologies and sufficient metadata: GSE139078, Sala Study, and PRJEB9292. Even so, experimental designs were not identical, and we identified significant batch effects that would have complicated functional analysis. Here we present quantile discretization and Bayesian network construction using the Hill climb method as a powerful tool to overcome experimental differences and reveal biologically relevant responses to the CF genotype itself, exposure to virus, bacteria, and drugs used to treat CF. Functional patterns revealed by cluster Profiler included interferon signaling, interferon gamma signaling, interleukins 4 and 13 signaling, interleukin 6 signaling, interleukin 21 signaling, and inactivation of CSF3/G-CSF signaling pathways showing significant alterations. These pathways were consistently associated with higher gene expression in CF epithelial cells compared to non-CF cells, suggesting that targeting these pathways could improve clinical outcomes. The success of quantile discretization and Bayesian network analysis in the context of CF suggests that these approaches might be applicable to other contexts where exactly comparable data sets are hard to find.

Background: Physician participation in clinical trials is essential for the progress of modern medicine. However, the demand for physician research partners is outpacing physicians' interest in participating in scientific studies. Understanding the factors that influence physician participation in research is crucial to addressing this gap., Methods: In this study, we used a physician's social network, as constructed from patient billing data, to study if the research choices of a physician's immediate peers influence their likelihood to participate in scientific research. We analyzed data from 348 physicians across 40 hospitals. We used logistic regression models to examine the relationship between a physician's participation in clinical trials and the participation of their social network peers, adjusting for age, years of employment, and influences from other hospital facilities., Results: We found that the likelihood of a physician participating in clinical trials increased dramatically with the proportion of their social network-defined colleagues at their primary hospital who were participating ([Formula: see text] for a 1% increase in the proportion of participating peers, [Formula: see text]). Additionally, physicians who work regularly at multiple facilities were more likely to participate ([Formula: see text], [Formula: see text]) and increasingly so as the extent to which they have social network ties to colleagues at hospitals other than their primary hospital increases ([Formula: see text], [Formula: see text]). These findings suggest an inter-hospital peer participation process., Conclusion: Our study provides evidence that the social structure of a physician's work-life is associated with their decision to participate in scientific research. The results suggest that interventions aimed at increasing physician participation in clinical trials could leverage the social networks of physicians to encourage participation. By identifying factors that influence physician participation in research, we can work towards closing the gap between the demand for physician research partners and the number of physicians willing to participate in scientific studies., (© 2023. The Author(s).)

Background: Transcriptomic profiling of adults with tuberculosis (TB) has become increasingly common, predominantly for diagnostic and risk prediction purposes. However, few studies have evaluated signatures in children, particularly in identifying those at risk for developing TB disease. We investigated the relationship between gene expression obtained from umbilical cord blood and both tuberculin skin test conversion and incident TB disease through the first 5 years of life., Methods: We conducted a nested case-control study in the Drakenstein Child Health Study, a longitudinal, population-based birth cohort in South Africa. We applied transcriptome-wide screens to umbilical cord blood samples from neonates born to a subset of selected mothers (N = 131). Signatures identifying tuberculin conversion and risk of subsequent TB disease were identified from genome-wide analysis of RNA expression., Results: Gene expression signatures revealed clear differences predictive of tuberculin conversion (n = 26) and TB disease (n = 10); 114 genes were associated with tuberculin conversion and 30 genes were associated with the progression to TB disease among children with early infection. Coexpression network analysis revealed 6 modules associated with risk of TB infection or disease, including a module associated with neutrophil activation in immune response (P < .0001) and defense response to bacterium (P < .0001)., Conclusions: These findings suggest multiple detectable differences in gene expression at birth that were associated with risk of TB infection or disease throughout early childhood. Such measures may provide novel insights into TB pathogenesis and susceptibility., Competing Interests: Potential conflicts of interest. D. J. S. reports grants or contracts from the United Kingdom MRC; royalties or licenses from American Psychiatric Press, Cambridge University Press, and Elsevier/Academic Press; consulting fees (paid to author) from Discovery Vitality, Kanna, Lundbeck, Orion, Sanofi, and Vistagen; payment or honoraria (directly to author) from Johnson & Johnson, L’Oreal, Servier, and Takeda; meeting and/or travel support from Lundbeck and Servier; and a role on a board, society, or committee for the African College of Neuropsychopharmacology. J. E. H. reports grants from the NIH, Bill & Melinda Gates Foundation, Cystic Fibrosis Foundation, National Sciences and Engineering Research Council of Canada, Australian Research Council, and Canada Research Chair Program; and received nonfinancial support (equipment, materials, or drugs) from Shimadzu and Markes International. C. A. B. reports consulting fees from ModernaTX (paid to institution for unrelated work). All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed., (© The Author(s) 2023. Published by Oxford University Press on behalf of Infectious Diseases Society of America.)

Protecting medical privacy can create obstacles in the analysis and distribution of healthcare graphs and statistical inferences accompanying them. We pose a graph simulation model which generates networks using degree and property augmentation and provide a flexible R package that allows users to create graphs that preserve vertex attribute relationships and approximating the retention of topological properties observed in the original graph (e.g., community structure). We illustrate our proposed algorithm using a case study based on Zachary's karate network and a patient-sharing graph generated from Medicare claims data in 2019. In both cases, we find that community structure is preserved, and normalized root mean square error between cumulative distributions of the degrees across the generated and the original graphs is low (0.0508 and 0.0514 respectively)., Competing Interests: Competing interestsThe authors declare that they have no competing interests., (© The Author(s) 2023.)

Preexisting antibodies to endemic coronaviruses (CoV) that cross-react with SARS-CoV-2 have the potential to influence the antibody response to COVID-19 vaccination and infection for better or worse. In this observational study of mucosal and systemic humoral immunity in acutely infected, convalescent, and vaccinated subjects, we tested for cross-reactivity against endemic CoV spike (S) protein at subdomain resolution. Elevated responses, particularly to the β-CoV OC43, were observed in all natural infection cohorts tested and were correlated with the response to SARS-CoV-2. The kinetics of this response and isotypes involved suggest that infection boosts preexisting antibody lineages raised against prior endemic CoV exposure that cross-react. While further research is needed to discern whether this recalled response is desirable or detrimental, the boosted antibodies principally targeted the better-conserved S2 subdomain of the viral spike and were not associated with neutralization activity. In contrast, vaccination with a stabilized spike mRNA vaccine did not robustly boost cross-reactive antibodies, suggesting differing antigenicity and immunogenicity. In sum, this study provides evidence that antibodies targeting endemic CoV are robustly boosted in response to SARS-CoV-2 infection but not to vaccination with stabilized S, and that depending on conformation or other factors, the S2 subdomain of the spike protein triggers a rapidly recalled, IgG-dominated response that lacks neutralization activity., Competing Interests: AC, HN, AH, CB, JW, WW, JL, EB, AT, AR, JB, DW, TG, AM, RC, PW, MA No competing interests declared

Background: Tuberculosis (TB) is the archetypical chronic infection, with patients having months of symptoms before diagnosis. In the two years after successful therapy, survivors of TB have a three-fold increased risk of death., Methods: Guinea pigs were infected with Mycobacterium tuberculosis ( Mtb ) for 45 days, followed by RRBS DNA methylation analysis. In humans, network analysis of differentially expressed genes across three TB cohorts were visualized at the pathway-level. Serum levels of inflammation were measured by ELISA. Horvath (DNA methylation) and RNA-seq biological clocks were used to investigate shifts in chronological age among humans with TB., Results: Guinea pigs with TB demonstrated DNA hypermethylation and showed system-level similarity to humans with TB ( p -value = 0.002). The transcriptome in TB in multiple cohorts was enriched for DNA methylation and cellular senescence. Senescence associated proteins CXCL9, CXCL10, and TNF were elevated in TB patients compared to healthy controls. Humans with TB demonstrate 12.7 years (95% CI: 7.5, 21.9) and 14.38 years (95% CI: 10.23-18.53) of cellular aging as measured by epigenetic and gene expression based cellular clocks, respectively., Conclusions: In both guinea pigs and humans, TB perturbs epigenetic processes, promoting premature cellular aging and inflammation, a plausible means to explain the long-term detrimental health outcomes after TB.

Scientific innovation has long been heralded the collaborative effort of many people, groups, and studies to drive forward research. However, the traditional peer review process relies on reviewers acting in a silo to critically judge research. As research becomes more cross-disciplinary, finding reviewers with appropriate expertise to provide feedback on an entire paper is increasingly difficult. We sought to pilot a crowd peer review process that allowed reviewers to interact with one another in the spirit of collaborative science. We focused this session on manuscripts using meta-analysis, to fully embrace the importance of collaborative and open scientific research in the field of biocomputing. Our pilot study found that researchers enjoy a more collaborative peer review process and felt that the process led to higher quality feedback for submitting authors than traditional review offers.

Spatially resolved characterization of the transcriptome and proteome promises to provide further clarity on cancer pathogenesis and etiology, which may inform future clinical practice through classifier development for clinical outcomes. However, batch effects may potentially obscure the ability of machine learning methods to derive complex associations within spatial omics data. Profiling thirty-five stage three colon cancer patients using the GeoMX Digital Spatial Profiler, we found that mixed-effects machine learning (MEML) methods† may provide utility for overcoming significant batch effects to communicate key and complex disease associations from spatial information. These results point to further exploration and application of MEML methods within the spatial omics algorithm development life cycle for clinical deployment.

Pre-existing antibodies to endemic coronaviruses (CoV) that cross-react with SARS-CoV-2 have the potential to influence the antibody response to COVID-19 vaccination and infection for better or worse. In this observational study of mucosal and systemic humoral immunity in acutely infected, convalescent, and vaccinated subjects, we tested for cross reactivity against endemic CoV spike (S) protein at subdomain resolution. Elevated responses, particularly to the β-CoV OC43, were observed in all natural infection cohorts tested and were correlated with the response to SARS-CoV-2. The kinetics of this response and isotypes involved suggest that infection boosts preexisting antibody lineages raised against prior endemic CoV exposure that cross react. While further research is needed to discern whether this recalled response is desirable or detrimental, the boosted antibodies principally targeted the better conserved S2 subdomain of the viral spike and were not associated with neutralization activity. In contrast, vaccination with a stabilized spike mRNA vaccine did not robustly boost cross-reactive antibodies, suggesting differing antigenicity and immunogenicity. In sum, this study provides evidence that antibodies targeting endemic CoV are robustly boosted in response to SARS-CoV-2 infection but not to vaccination with stabilized S, and that depending on conformation or other factors, the S2 subdomain of the spike protein triggers a rapidly recalled, IgG-dominated response that lacks neutralization activity.

Non-alcoholic steatohepatitis (NASH) is a fatty liver disease characterized by accumulation of fat in hepatocytes with concurrent inflammation and is associated with morbidity, cirrhosis and liver failure. After extraction of a liver core biopsy, tissue sections are stained with hematoxylin and eosin (H&E) to grade NASH activity, and stained with trichrome to stage fibrosis. Methods to computationally transform one stain into another on digital whole slide images (WSI) can lessen the need for additional physical staining besides H&E, reducing personnel, equipment, and time costs. Generative adversarial networks (GAN) have shown promise for virtual staining of tissue. We conducted a large-scale validation study of the viability of GANs for H&E to trichrome conversion on WSI (n = 574). Pathologists were largely unable to distinguish real images from virtual/synthetic images given a set of twelve Turing Tests. We report high correlation between staging of real and virtual stains ([Formula: see text]; 95% CI: 0.84-0.88). Stages assigned to both virtual and real stains correlated similarly with a number of clinical biomarkers and progression to End Stage Liver Disease (Hazard Ratio HR = 2.06, 95% CI: 1.36-3.12, p < 0.001 for real stains; HR = 2.02, 95% CI: 1.40-2.92, p < 0.001 for virtual stains). Our results demonstrate that virtual trichrome technologies may offer a software solution that can be employed in the clinical setting as a diagnostic decision aid.

Pediatric tuberculosis (TB) remains a global health crisis. Despite progress, pediatric patients remain difficult to diagnose, with approximately half of all childhood TB patients lacking bacterial confirmation. In this pilot study (n = 31), we identify a 4-compound breathprint and subsequent machine learning model that accurately classifies children with confirmed TB (n = 10) from children with another lower respiratory tract infection (LRTI) (n = 10) with a sensitivity of 80% and specificity of 100% observed across cross validation folds. Importantly, we demonstrate that the breathprint identified an additional nine of eleven patients who had unconfirmed clinical TB and whose symptoms improved while treated for TB. While more work is necessary to validate the utility of using patient breath to diagnose pediatric TB, it shows promise as a triage instrument or paired as part of an aggregate diagnostic scheme.

Understanding humoral immune responses to SARS-CoV-2 infection will play a critical role in the development of vaccines and antibody-based interventions. We report systemic and mucosal antibody responses in convalescent individuals who experienced varying severity of disease. Whereas assessment of neutralization and antibody-mediated effector functions revealed polyfunctional antibody responses in serum, only robust neutralization and phagocytosis were apparent in nasal wash samples. Serum neutralization and effector functions correlated with systemic SARS-CoV-2-specific IgG response magnitude, while mucosal neutralization was associated with nasal SARS-CoV-2-specific IgA. Antibody depletion experiments support the mechanistic relevance of these correlations. Associations between nasal IgA responses, virus neutralization at the mucosa, and less severe disease suggest the importance of assessing mucosal immunity in larger natural infection cohorts. Further characterization of antibody responses at the portal of entry may define their ability to contribute to protection from infection or reduced risk of hospitalization, informing public health assessment strategies and vaccine development efforts., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2021 Butler, Crowley, Natarajan, Xu, Weiner, Bobak, Mattox, Lee, Wieland-Alter, Connor, Wright and Ackerman.)

Research into AI implementations for healthcare continues to boom. However, successfully launching these implementations into healthcare clinics requires the co-operation and collaboration of multiple stakeholders in healthcare including healthcare professionals, administrators, insurers, legislators, advocacy groups, as well as the patients themselves. The co-operation and collaboration of these interprofessional groups is necessary not just in the final stages of launching AI based solutions in healthcare, but along each stage of the research design and analysis. In this workshop, we solicited talks from researchers who have embraced the idea of interprofessional collaboration across many different stakeholder groups at multiple stages of their research. We specifically focus on projects which included heavy collaborations from healthcare professionals, embraced the research subjects' communities as critical research partners, as well as included researchers who are advocating for systemized changes to include interprofessional stakeholders as evaluators of AI research in healthcare.

The growth of publicly available repositories, such as the Gene Expression Omnibus, has allowed researchers to conduct meta-analysis of gene expression data across distinct cohorts. In this work, we assess eight imputation methods for their ability to impute gene expression data when values are missing across an entire cohort of Tuberculosis (TB) patients. We investigate how varying proportions of missing data (across 10%, 20%, and 30% of patient samples) influence the imputation results, and test for significantly differentially expressed genes and enriched pathways in patients with active TB. Our results indicate that truncating to common genes observed across cohorts, which is the current method used by researchers, results in the exclusion of important biology and suggest that LASSO and LLS imputation methodologies can reasonably impute genes across cohorts when total missingness rates are below 20%.

Background: Intraclass correlation coefficients (ICC) are recommended for the assessment of the reliability of measurement scales. However, the ICC is subject to a variety of statistical assumptions such as normality and stable variance, which are rarely considered in health applications., Methods: A Bayesian approach using hierarchical regression and variance-function modeling is proposed to estimate the ICC with emphasis on accounting for heterogeneous variances across a measurement scale. As an application, we review the implementation of using an ICC to evaluate the reliability of Observer OPTION 5 , an instrument which used trained raters to evaluate the level of Shared Decision Making between clinicians and patients. The study used two raters to evaluate recordings of 311 clinical encounters across three studies to evaluate the impact of using a Personal Decision Aid over usual care. We particularly focus on deriving an estimate for the ICC when multiple studies are being considered as part of the data., Results: The results demonstrate that ICC varies substantially across studies and patient-physician encounters within studies. Using the new framework we developed, the study-specific ICCs were estimated to be 0.821, 0.295, and 0.644. If the within- and between-encounter variances were assumed to be the same across studies, the estimated within-study ICC was 0.609. If heteroscedasticity is not properly adjusted for, the within-study ICC estimate was inflated to be as high as 0.640. Finally, if the data were pooled across studies without accounting for the variability between studies then ICC estimates were further inflated by approximately 0.02 while formerly allowing for between study variation in the ICC inflated its estimated value by approximately 0.066 to 0.072 depending on the model., Conclusion: We demonstrated that misuse of the ICC statistics under common assumption violations leads to misleading and likely inflated estimates of interrater reliability. A statistical analysis that overcomes these violations by expanding the standard statistical model to account for them leads to estimates that are a better reflection of a measurement scale's reliability while maintaining ease of interpretation. Bayesian methods are particularly well suited to estimating the expanded statistical model.