27941 research outputs found
Sort by
Decoding cancer prognosis with deep learning: the ASD-cancer framework for tumor microenvironment analysis
No full textDeep learning is revolutionizing biomedical research by facilitating the integration of multi-omics data sets while bridging classical bioinformatics with existing knowledge. Building on this powerful potential, Zhang et al. proposed a semi-supervised learning framework called Autoencoder-Based Subtypes Detector for Cancer (ASD-cancer) to improve the multi-omics data analysis (H. Zhang, X. Xiong, M. Cheng, et al., 2024, mSystems 9:e01395-24, https://doi.org/10.1128/msystems.01395-24). By utilizing autoencoders pre-trained on The Cancer Genome Atlas data, the ASD-cancer framework outperforms the baseline model. This approach also makes the framework scalable, enabling it to process new data sets through transfer learning without retraining. This commentary explores the methodological innovations and scalability of ASD-cancer while suggesting future directions, such as the incorporation of additional data layers and the development of adaptive AI models through continuous learning. Notably, integrating large language models into ASD-cancer could enhance its interpretability, providing more profound insights into oncological research and increasing its influence in cancer subtyping and further analysis.No embarg
A Culturally Tailored Artificial Intelligence Chatbot (K-Bot) to Promote Human Papillomavirus Vaccination Among Korean Americans: Development and Usability Study
No full textBackground: Human papillomavirus (HPV) is the most common sexually transmitted infection (STI) worldwide and is associated with various cancers, including cervical and oropharyngeal cancers. Despite the availability of effective vaccines, significant disparities in HPV vaccination rates persist, particularly among racial and ethnic minorities, such as Korean Americans. Cultural stigma, language barriers, and limited access to tailored health information contribute to these disparities.
Objective: This study aimed to develop and evaluate the usability of K-Bot, an artificial intelligence (AI)-powered, culturally tailored, bilingual (Korean and English) chatbot designed to provide culturally sensitive health information about HPV vaccination to Korean immigrants and Korean Americans.
Methods: K-Bot was developed using CloudTuring and Google Dialogflow. Its dialogues were created using Centers for Disease Control and Prevention (CDC) evidence-based HPV information and tailored to the Korean American population based on findings from previous studies. The evaluation and refinement process for K-Bot was organized into 3 phases: (1) expert evaluation by a multidisciplinary panel, (2) usability testing, and (3) iterative refinement based on feedback. An online survey collected demographics, HPV awareness, and vaccination status before 6 focus groups (N=21) sessions using semistructured questions guided by Peter Morville's usability framework. Quantitative data were analyzed descriptively, and thematic analysis assessed usability, cultural relevance, and content clarity across 6 dimensions: desirability, accessibility, findability, credibility, usability, and usefulness.
Results: Participants had a mean age of 23.7 (SD 4.7) years, with most being female (n=12, 57.1%), second-generation individuals (n=13, 61.9%), and single (n=20, 95.2%). HPV awareness was high (n=19, 90.5%), vaccine knowledge was also high (n=18, 81.8%), but only 11 (52.4%) participants were vaccinated. Feedback-driven refinements addressed usability challenges, including simplifying navigation and adding visual elements. Participants described K-Bot as a promising tool for promoting HPV vaccination among Korean and Korean American users, citing its bilingual functionality and culturally tailored content as key strengths. Evidence-based information was valued, but participants recommended visuals to improve engagement and reduce cognitive load. Accessibility concerns included broken links, and participants proposed enhancements, such as animations, demographic-specific resources, and interactive features, to improve usability and engagement further.
Conclusions: Usability testing of K-Bot revealed its potential as a culturally tailored, bilingual tool for promoting HPV vaccination among Korean immigrants and Korean Americans. Participants valued its evidence-based information, cultural relevance, and bilingual functionality but recommended improvements, such as enhanced navigation, visual elements, and interactive features, to boost engagement and usability. These findings support the potential of AI-driven tools to improve health care access by addressing key barriers to care. Further research is needed to evaluate their broader impact and optimize their design and implementation for individuals with diverse health care needs.No embarg
Thalamic Nuclear Volumes in Fetal Alcohol Spectrum Disorders: from Adolescence to Middle-Age 20 Years Later
No full textBackground: Midline orofacial and brain structures, including the multinucleated thalamus, may be differentially sensitive to prenatal alcohol exposure and vulnerable to accelerated aging.
Methods: Two sets of MRI data separated by 20 years are reported for controls, individuals with fetal alcohol syndrome (FAS), and nondysmorphic individuals with heavy fetal alcohol exposure (FAE). MRI1 included 179 participants with 69 reassessed at MRI2. Segmentation produced estimates of bilateral thalamic volume and 10 bilateral nuclei, which were aggregated into Anterior, Ventral, Posterior, and Medial Volumes. Differences were assessed without and with correction for intracranial volume (ICV).
Results: MRI1 revealed stepwise group differences in ICV, total thalamic volume, and Anterior and Ventral regions uncorrected for ICV, where Controls>FAE>FAS. Corrected for ICV, the smaller volumes endured in the Anterior and Ventral regions, although differences between FAE and FAS groups were attenuated. Nuclei volumes were selectively smaller in the alcohol-exposed groups than controls even after controlling for ICV. Longitudinally, thalamic volumes typically declined over time maintaining the stepwise effects and with little evidence for accelerated decline in the FAE or FAS groups.
Conclusions: These novel data revealed stable deficits in thalamic nuclei of the groups with heavy fetal alcohol exposure. After 20 years, the deficits endured but without accelerated age-related decline and following the same aging pattern as controls. Despite parallel aging functions in all groups, ICV adjustment yielded volume deficits localized to the anterior and ventral thalamic nuclei, differing from patterns in the remaining thalamic nuclei and cortical brain structures.No embarg
Hope for the Future: Key Informants' Perspectives on HIV Prevention in Dominican Republic Batey Communities: A Qualitative Description Study
No full textThe purpose of our study was to understand the interlocking spheres of cultural identity and health behaviors related to HIV prevention within Haitian migrant batey communities in La Romana, Dominican Republic. A qualitative description design was employed using the PEN-3 model by Airhihenbuwa (1990) as a theoretical framework. Data were collected through semi-structured interviews with 12 key informants. Participants, primarily adults of Hispanic (10) and African descent (2), ranged from 33 to 53 years old, with a majority having high school or higher education. A central theme, "Hope for the Future," emerged, highlighting five subthemes: stigma/discrimination, religious beliefs, voodoo tenets, community nurturers, and HIV education. Findings emphasized the need for a multifaceted approach using community health workers and incorporating local cultural contexts, including religious beliefs and stigma, to enhance HIV prevention efforts. Cultural identity of Haitian migrant batey communities included religious beliefs, stigma, and cultural practices as considerations in HIV prevention interventions.No embarg
Machine Learning and Artificial Intelligence Approaches for the Design of Potent Therapeutic siRNAs
No full textSmall interfering RNAs (siRNAs) represent a transformative class of therapeutics, with the potential to target previously undruggable genes. However, identifying potent silencing siRNAs remains a challenge, particularly in the context of fully chemically modified scaffolds required for therapeutic applications. This dissertation addresses these challenges by developing, validating, and deploying the first AI-powered design platform specifically optimized for the development of fully modified therapeutic siRNAs.
To build toward this solution, we systematically developed and evaluated siRNA efficacy prediction models across a series of frameworks including linear models, supervised and semi-supervised machine learning, and deep learning-based sequence feature encoding strategies. We assembled what will be the largest (∼5,000) publicly available dataset of fully modified endogenously evaluated siRNAs and developed a semi-supervised learning model that achieved unprecedented predictive performance (F1-score ∼0.7) supported by experimental validation. Drawing from these models by conducting explainable AI techniques, we uncover mechanistic insights driving predictions that agree with the current consensus on siRNA mechanisms. We explore the impacts of chemical modifications and assay context on siRNA efficacy. Through semi-supervised modeling we demonstrate that human-trained models fail to generalize to mouse models, highlighting a potential role of sequence context in siRNA targeting.
To ensure accessibility, we deployed our model as a user-friendly, publicly available web tool for therapeutic siRNA design (https://kmonopoli.github.io/sirna). This work not only advances the field of siRNA prediction but also establishes a new paradigm for AI-driven oligonucleotide drug discovery, offering researchers a robust, validated platform for rapid preclinical candidate selection.Interdisciplinary Graduate Program2 years2027-09-1
Investigating how the HIV Protease Inhibitor, Nelfinavir Enhances MHC I Antigen Presentation through the ER-associated degradation (ERAD) Pathway
No full textMHC-I antigen presentation is a critical surveillance mechanism used by the immune system to monitor the status of cells by generating and displaying on the cell surface peptides derived from the proteome. Antigenic peptides such as those generated by transformed cells and viruses would get presented on the cell surface in the form of peptide: MHC I complex for recognition and killing by CD8 T-cells. Methods to increase MHC I antigen presentation could enhance such immunity. Moreover, to avoid detection by CD8 T-cells, cancers and viruses have developed many evasion strategies to inhibit various steps in the MHC I antigen presentation pathway and methods to counteract these immune evasion mechanisms could be useful. Therefore, it is important to understand how MHC I levels are controlled and the strategies that can be used to enhance the expression of MHC I. In this dissertation, I will discuss how the HIV protease inhibitor, nelfinavir can enhance surface MHC I expression in various cell lines including cancer cells by blocking the ability of MHC I to undergo the ER associated degradation pathway (ERAD) and therefore, preventing MHC I molecules from translocating from the ER to cytosol for proteasomal degradation.Immunology and Microbiology2 years2026-05-0
T Cell Lymphopenia in Severe Congenital Neutropenia Type 5 Mouse Model
No full textVPS45 is a regulator of membrane trafficking. Severe Congenital Neutropenia Type 5 (SCN5) is associated with mutations in VPS45. VPS45 is ubiquitously expressed, while SCN5 is characterized by a reduction in neutrophils. The focus of this work is to discover the link between VPS45 and SCN5. In collaboration with the Newburger lab, a mouse model of the SCN5 mutation was created. Initial discoveries of the mouse model parallel the SCN5 phenotype, failure to thrive, with lower weights and a smaller appearance. It was discovered that T cell lymphopenia was the primary effect of these VPS45 KI mutations. The VPS45 mutant T cells have a more activated phenotype. The development of the VPS45 mutant T cells were abnormal, lacking some of the initial stages of development in the thymus. This work, although preliminary, illuminates the importance of VPS45 and proper membrane trafficking in immune cells.Biochemistry and Molecular Biotechnology2 years2027-04-1
Circulating extra-cellular RNAs and atrial fibrillation: data from the TRACE-CORE cohort
Get PDFBackground: Atrial fibrillation (AF) is the most common sustained arrhythmia and is linked to increased risk of stroke, heart failure, and mortality. Circulating extracellular RNAs (exRNAs), which regulate gene expression and reflect underlying biological processes, are potential biomarkers for atrial fibrillation.
Methods: As part of an ongoing, larger study into extracellular RNAs (exRNAs) as potential biomarkers for cardiovascular disease, we analyzed exRNA profiles in a subset of 296 survivors of acute coronary syndrome (ACS) enrolled in the Transitions, Risks, and Actions in Coronary Events Center for Outcomes Research and Education (TRACE-CORE) cohort. A total of 318 exRNAs were quantified, selected a priori based on prior findings from the Framingham Heart Study. We assessed associations between circulating exRNAs and echocardiographic intermediate phenotypes relevant to atrial fibrillation (AF), including left atrial dimension, left ventricular (LV) mass, LV end-diastolic volume, and global longitudinal strain. Subsequently, we used logistic regression models to evaluate whether the exRNAs associated with these phenotypes were also associated with a history of AF (n = 18, 5.4%). Downstream bioinformatics analyses were performed to identify putative target genes, enriched gene ontology categories, and molecular pathways regulated by these candidate microRNAs.
Results: We identified 77 extracellular RNAs (exRNAs) that were significantly associated with increased left ventricular (LV) mass and at least one additional echocardiographic intermediate phenotype. Among these, miR-17-5p and miR-574-3p were also significantly associated with a history of atrial fibrillation (AF), with odds ratios of 1.58 (95% CI: 1.10-2.26) and 2.16 (95% CI: 1.03-4.54), respectively. Predicted gene targets of these miRNAs were enriched in pathways implicated in atrial remodeling and arrhythmogenesis. Key overlapping canonical pathways included the Senescence Pathway, Idiopathic Pulmonary Fibrosis Signaling, ERK5 Signaling, RHO GTPase Cycle, and HGF Signaling.
Conclusions: Circulating exRNAs, including miR-17-5p and miR-574-3p, are associated with cardiac remodeling and a history of AF in ACS survivors. These findings highlight their potential as biomarkers of atrial remodeling and implicate key molecular pathways involved in AF pathogenesis.No embarg
Causal Inference via Electronic Health Records in the National Clinical Cohort Collaborative: Challenges and Solutions in Long COVID Research [preprint]
No full textThis article is a preprint. Preprints are preliminary reports of work that have not been certified by peer review.Observational analyses of electronic health record (EHR) data using databases such as the National Clinical Cohort Collaborative include unique challenges for researchers seeking causal inferences, particularly when evaluating subjectively-defined outcomes like Long COVID. We explore several challenges and describe potential solutions. 1. Lack of true negatives: Many diagnoses and conditions either have a positive indicator or a missing status, requiring investigators to carefully consider which patients are likely negative for this condition. 2. Differential monitoring: EHR data include nonrandom missingness driven by patients engaging with the healthcare system at different rates, which is often related to both the exposure and outcome of interest. 3. Bias: EHR data sources face many biases, but are particularly vulnerable to informative missingness, differential monitoring, and model misspecification. 4. Large sample size: High precision (i.e., narrow confidence intervals) paired with potential bias leads to a high risk of incorrectly rejecting the null hypothesis. 5. Defining index time: It is important that investigators deliberately define index time (i.e., , baseline) to ensure that they only adjust for baseline confounders and do not adjust for (or condition on) factors that are affected by the exposure of interest (i.e., colliders or mediators). 6. Parameter selection: Investigators should only select parameters that are supported by the data distribution. This manuscript provides an overview of these challenges and solutions, using both simulated data and real-world data, with the outcome of Long COVID as the running example.The UMass Center for Clinical and Translational Science (UMCCTS), UL1TR001453, helped fund this study.No embarg
Improving Women Veterans’ Experience of Obstetric Care in the United States: Findings From a Mixed-Methods Analysis of VA-Purchased Maternity Care
Get PDFIntroduction: Women Veterans represent a growing and medically complex population within the U.S. healthcare system, yet their experiences with maternity care—especially when delivered outside the VA through Community Care Network (CCN) providers—remain under-examined. This dissertation: (1) investigates racial and ethnic disparities in obstetric outcomes, (2) evaluates a community-based doula intervention, and (3) explores how Black, Indigenous, and People of Color (BIPOC) Veterans navigate provider selection and engage with perinatal care (PC) quality measures.
Methods: Drawing on data from the Center for Maternal and Infant Outcomes Research in Translation (COMFORT) study—a multi-site national cohort of pregnant Veterans—three analyses were conducted: (1) a retrospective cohort analysis assessing racial and ethnic differences in unplanned cesarean birth among 314 primiparous Veterans with term, singleton deliveries; (2) a mixed-methods pilot study evaluating the feasibility and acceptability of doula care among 39 referred Veterans; and (3) a focused analysis of 27 BIPOC Veterans describing their provider selection and prioritization of three PC quality measures.
Results: Non-Hispanic BIPOC Veterans had significantly higher risk of unplanned cesarean birth than White Veterans, even after adjusting for clinical risk. Doula care seemed feasible and acceptable as a potential VA maternity care benefit, with participants expressing strong interest in future use. Most Veterans selected providers based on geographic proximity or CCN inclusion. Although participants were largely unaware of obstetric quality measures or how to access and interpret such data prior to the study, when asked to rate three PC quality measures, they consistently prioritized infant safety—rating “unexpected complications in term newborns” as most important.
Conclusion: This dissertation identifies disparities and structural barriers in VA-purchased maternity care and underscores the need for improved oversight of the CCN, culturally responsive support models, and Veteran-centered tools for informed perinatal decision-making.MD/PhDPopulation Health SciencesNo embarg