95038 research outputs found
Sort by
AI‐Driven Variant Annotation for Precision Oncology in Breast Cancer
Get PDFInterpreting the functional impact of genomic variants remains a major challenge in precision oncology, particularly in breast cancer, where many variants of unknown significance lack clear therapeutic guidance. Current annotation strategies focus on frequent driver mutations, leaving rare or understudied variants unclassified and clinically uninformative. Here, we present an Artificial Intelligence/Machine Learning (AI/ML)-driven framework that systematically identifies variants associated with key breast cancer phenotypes, including ESR1 and EZH2 activity, by integrating genomic, transcriptomic, structural, and drug response data. Using CCLE/DepMap and TCGA datasets, we analyzed > 12,000 variants across breast cancer genomes, identifying structurally clustered mutations that share functional consequences with well-characterized oncogenic drivers. This approach reveals that mutations in PIK3CA, TP53, and other genes strongly associate with ESR1 signaling, challenging conventional assumptions about endocrine therapy response. Additionally, EZH2-associated variants emerge in unexpected genomic contexts, suggesting new targets for epigenetic therapies. By shifting from frequency-based to structure-informed classification, we expand the set of potentially actionable mutations, enabling improved patient stratification and drug repurposing strategies. This work provides a scalable, clinically relevant method to accelerate variant annotation, offering new insights into drug sensitivity and resistance mechanisms. Future validation efforts will refine these predictions and integrate clinical outcomes to guide personalized treatment strategies. Our findings highlight the transformative potential of AI/ML in redefining cancer variant interpretation, bridging the gap between genomics, functional biology, and precision medicine
Extreme Sign Reversals in Regression and Simpson’s Paradox
Get PDFWe congratulate the authors of Chatterjee et al. (2025) on a very interesting article introducing the uniLasso method, which is designed to preserve, with high probability, the sign of a predictor’s slope in simple regression during the variable selection process for the full model. Preserving the sign not only enhances the stability of model selection but, more importantly, improves the interpretability of the resulting model. This feature makes the method especially valuable in practice: it produces a sparse set of predictors while ensuring that the direction of each predictor’s relationship with the response remains reliable
Modeling Protein–Protein and Protein–Ligand Interactions by the ClusPro Team in CASP16
Get PDFIn the CASP16 experiment, our team employed hybrid computational strategies to predict both protein-protein and protein-ligand complex structures. For protein-protein docking, we combined physics-based sampling-using ClusPro FFT docking and molecular dynamics-with AlphaFold (AF)-based sampling, followed by AF-based refinement. Our method produced numerous high-accuracy complex models, including cases where AF alone failed, underscoring the critical role of physics-based sampling alongside deep learning-based refinement. For protein-ligand docking, we integrated the ClusPro LigTBM template-based approach with a machine learning-based confidence model for rescoring. The method preserves conserved interaction fragments derived from homologous complexes, followed by local resampling using physics-based sampling and a diffusion model. Our template-based strategy achieved a mean lDDT-PLI of 0.69 across 233 targets, which was highly competitive. These results demonstrate that combining physics-based modeling with AI-driven refinement can significantly enhance the accuracy of both protein-protein and protein-ligand structure predictions
Goldilocks Zone: Allergists' Perceptions of Family Anxiety and Investment in Food Allergy Oral Immunotherapy.
Get PDFOral immunotherapy (OIT) is an increasingly popular but still contentious treatment option for food allergies. Since most OIT patients are children, families' involvement in treatment is an important part of the discussion of the appropriateness of this treatment option for individual patients.To explore how US-based allergists judge whether OIT is a good fit for specific patients and their families.Providers were recruited through direct solicitation by email to participate in an in-depth interview about their perceptions of the risks and benefits of current and future food allergy treatment options, including OIT.60 interviews were conducted with academic and community providers from 34 states. The primary finding was that as part of the shared decision-making process for OIT, providers actively assessed families' levels of anxiety and investment in this treatment option. They were seeking levels of both that were "just right" and found families with both too low and too high levels of anxiety and investment as a poor fit for OIT.Providers used assessments of family anxiety and investment as mechanisms to optimize the benefits and minimize the risk of harm to their patients from OIT as well as to streamline staffing burdens in their practices
Variance Components, Correlation Components, Canonical Correlation, and Prediction in Mixed Models
Get PDFAnalyzing longitudinal outcomes presents distinct challenges due to within- and between-subject variation, manifesting as within-subject correlation. Linear mixed models (LMMs) and canonical correlation analysis (CCA) offer valuable insights into the relationships within such data structures. LMMs are fundamental for accommodating both fixed and random effects in longitudinal data analysis. CCA, traditionally applied to independent observations, explores relationships between two sets of variables via linear combinations. Combining these methodologies enhances understanding of both within-subject and between-subject associations in longitudinal data. Chapter 2 extends the concept of the intraclass correlation coefficient (ICC) to a broad class of linear and generalized linear mixed models. This extension introduces a variance fraction index that differs from pairwise correlations. A general formula links pairwise correlation and variance fraction indices, applicable to all mixed models, including those addressing overdispersion. These indices are demonstrated in linear, logistic, and loglinear mixed models, with graphical evaluations using real data. Chapter 3 investigates the estimation and inference of canonical correlation in the context of linear mixed models. Canonical correlation is used to interpret and summarize estimates of variance components, addressing issues such as temporal misalignment and missing values in longitudinal data. The canonical correlation parameter is estimated using variance-component estimates from linear mixed models. CCA is also examined as a tool for interpreting variance components and assessing covariate assumptions. Chapter 4 explores relationships between the random effects structure of linear mixed models and intracluster correlations frequently used to design and power cluster randomized trials. Methods to generate marginal pairwise correlations for mixed models, including continuous, binary, and count outcomes, are described. These pairwise correlations link mixed and marginal models and are functions of variance fractions explained by cluster factors. The approach is illustrated with a stepped wedge cluster randomized trial for antibiotic stewardship, analyzed using generalized estimating equations.Doctor of Public Healt
THREE ESSAYS ON ACCESS AND UTILIZATION OF HEALTHCARE
Get PDFThis dissertation examines three critical dimensions of healthcare access and utilization in the United States. The first chapter investigates the relationship between prenatal care adequacy and maternal morbidity outcomes using North Carolina birth records (2011-2019). Employing multivariate logistic regression, the analysis demonstrates that inadequate prenatal care increases the probability of maternal morbidity by 7.2% (OR=1.072, 95% CI: 1.01-1.13) in the general birthing population and by 45.6% (OR=1.456, 95% CI: 1.03-2.07) among individuals with preexisting diabetes and/or hypertension. These findings underscore the protective role of adequate prenatal care, particularly for vulnerable populations with preexisting conditions.The second chapter addresses primary care shortages by examining how nurse practitioners' scope of practice (SOP) legislation affects healthcare utilization. Using Medicare claims data and a difference-in-difference approach with heterogeneous treatment effects, the research reveals that expanding nurse practitioners' SOP increases evaluation and management service utilization by approximately 155 services per 10,000 Medicare beneficiaries (95% CI: -209.4-519.4). This suggests that policy interventions expanding nurse practitioners' authority may improve healthcare access for older adults. The findings indicate that future research should explore additional mechanisms through which SOP expansion affects healthcare access, particularly how nurse practitioners might facilitate access to specialty care through referrals. The third chapter explores non-financial barriers to healthcare utilization among older adults using 2019 Health Survey data. The analysis identifies that dislike of visiting doctors decreases the likelihood of attempting to seek care by 65.7% (95% CI: 0.21-0.56) compared to those without such aversion. Furthermore, significant disparities exist, with non-White respondents 42.8% (95% CI: 0.43-0.76) less likely and Hispanic respondents 46.2% (95% CI: 0.37-0.77) less likely to attempt seeking care compared to their White and non-Hispanic counterparts, respectively.Collectively, this research offers multifaceted insights into factors influencing healthcare access and utilization across different populations. By identifying the importance of prenatal care for maternal health, examining policy interventions to address provider shortages, and understanding non-financial barriers to care-seeking behavior, this dissertation contributes valuable evidence to inform healthcare policies aimed at improving access, reducing disparities, and enhancing health outcomes.Doctor of Philosoph
Engaging at-Risk Patients In a Safety Planning Intervention: A Program Evaluation of Zero Suicide in a Psychiatric Emergency Department
Get PDFSuicide remains a public health issue and suicide care needs to be an integral part ofhealthcare delivery. In studies looking at medical system use patterns of people who die bysuicide, evidence demonstrates that these patients do indeed seek medical and or psychiatriccare in the year before their death. Specific to emergency department (ED) visits, researchstudies found that 7% of all patients who died by suicide were seen at an ED within a week oftheir death, while 14% were seen within a month, and 43.8% within the previous year. ZeroSuicide, a framework for suicide prevention, is based on the premise that there is no acceptablenumber of suicides, and that the healthcare industry needs to address this issue with evidence-based practices to achieve this goal. This project aimed to conduct a program evaluation of thecurrent state of Zero Suicide implementation of safety planning in the Atrium Health PsychiatricED at Behavioral Health-Charlotte using the CDC Model. Assessment methods included chartreview of high and moderate-risk patients seen in the ED over 3 months, evaluation ofeducational offerings and staff adherence, a survey of knowledge and attitudes of current staffregarding ZS and safety planning, and comparison of current versus ideal state of EMRutilization to promote safety planning in the ED. Using data collected from March 1 to May 31,2024, 457 ED adult patients screened as either high or moderate risk for suicide on the Columbia Suicide Severity Rating Scale (C-SSRS) were discharged from either the ED orObservation (OBS) units. Safety planning, using the Stanley-Brown Suicide Safety Planning tool,was conducted with 22.5% of these patients. Discharge location was closely correlated withwhether a patient received a safety plan, with OBS patients having received a safety plan in75% of cases in contrast to just 1.2% of ED discharges. Findings from policy review, staffevaluation (survey and meetings/shadowing), EMR review, and educational assessmentsindicated multiple gaps in implementation and identified opportunities for improved suicidecare in the ED.Doctor of Nursing Practic
QUALITY IMPROVEMENT PROJECT FOR ADOLESCENT MENTAL HEALTH THROUGH TABLET-BASED SCREENING
Get PDFBACKGROUND: About 20% of adolescents have had one major depressive episode, with 59% of them not receiving treatment. About 32% of adolescents have an anxiety disorder, with 8% experiencing severe impact from it. Screening for adolescent depression and anxiety is recommended by the American Academy of Pediatrics and the United States Preventative Services Task Force (USPSTF). At the University of North Carolina (UNC) Adolescent Specialty Care Clinic (ASCC), screening is an already established process; however, a strategic goal was to transition paper-based screenings to tablet-based screenings (TBS).METHODS: This quality improvement (QI) initiative sought to transition the mental health screenings of the PHQ-SADS and the EAT-26 that were used in the ASCC from paper-based to TBS to provide benefits to the healthcare team using quality improvement methodologies, chart audits, and surveys to assess the success and benefits of TBS. Through standardization and electronic medical record integration (EMR) and modification, TBS was introduced. Quality improvement methodologies of key stakeholder interviews, process mapping, plan-do-study-act (PDSA) cycles, and real-time data feedback guided the implementation. Education using multiple methods, such as one-on-one training and tip sheets, improved adherence to the process. RESULTS: TBS occurred 50-91% of the time. Possible barriers to TBS included the legal limitations of the mental health screeners, limited tablets, and the change of process that interrupted the usual process flow. Survey results found staff to be favorable towards the change, with 100% of respondents being “definitely satisfied” with the process change and 80% believing it was useful, saved time, improved appointment flow, helped documentation to be more timely and complete, and will be a sustained change.CONCLUSION: TBS was found to be beneficial to healthcare providers to increase provider satisfaction around screening, decrease documentation burden, and streamline appointments. TBS may have issues related to limited resources or the legality of using electronic screening forms for sensitive screeners. TBS can not only address the growing adolescent mental health crisis but also address burdens providers experience in documentation burden.Doctor of Nursing Practic
QUEER, SICK, AND POOR: FINANCIAL HARDSHIP AMONG LGBTQ+ CANCER SURVIVORS AND CAREGIVERS
Get PDFSignificance: Lesbian, gay, bisexual, transgender and other sexual and gender minority (LGBTQ+) populations make up a substantial proportion of the United States (US) population—over 23 million individuals. LGBTQ+ populations throughout the US face identity-related stigma, ranging from interpersonal to structural, that leads to a variety of physical, mental, and economic health disparities. At the same time, the rising costs of cancer care puts over half of cancer survivors at risk of financial hardship and the subsequent outcomes—poor quality of life and reduced access to care. Emerging research suggests that LGBTQ+ cancer survivors and caregivers may experience worse financial hardship in comparison to non-LGBTQ+ counterparts, driven by existing economic inequities and multi-level anti-LGBTQ+ stigma. However, inequity in financial burden has not been estimated using national datasets, nor, has an intersectional lens been taken to interrogate the diversity of the LGBTQ+ population. Innovation: Employing multiple national datasets with sexual orientation and gender identity data (SOGI) data (Aims 1 and 2) as well as an explanatory sequential mixed-methods design with in-depth qualitative interviews (Aim 3) this dissertation provided a novel opportunity to provide national findings among LGBTQ+ cancer survivors paired with in-depth experiences of LGBTQ+ cancer caregivers. Integration of cancer-related financial hardship frameworks and the theory of intersectionality is not only responsive to national calls to center health equity (NOT-HS-21-014), but it also allows, for the first time, comprehensive and robust estimates to be generated for LGBTQ+ cancer survivors of diverse identities. Aims: Aim 1 used multivariable logit regression to estimate inequities in material, behavioral, and psychological financial hardship between LGB (e.g., lesbian, gay, bisexual) and non-LGB cancer survivors in the National Health Interview Survey. Aim 2 employed intersectional analyses to estimate inequity in behavioral financial hardship across identities and ages between LGBTQ+ and non-LGBTQ+ cancer survivors in the All of Us Research Program dataset. Aim 3 used an explanatory sequential mixed-methods design including in-depth qualitative interview building on an existing federally funded parent grant study. Aim 3 analyses described the lived experience of financial burden among LGBTQ+ cancer caregivers as well as provided multi-level LGBTQ+ specific recommendations. Study Impact: This study employs multiple data sources that each provide unique and important context to LGBTQ+ inequities in financial hardship. Affordability is crucial to access to care, thus understanding the between and within LGBTQ+ community variation in financial hardship promotes equity in health services delivery. Overall, the purpose of this study was to estimate inequity, identify modifiable factors, and elicit LGBTQ+ recommendations to improve the provision of cancer health services to LGBTQ+ populations.Doctor of Philosoph
STANDARDIZING ORAL ANTICANCER MEDICATION INITIATION TO IMPROVE CLINICIAN COMPLIANCE WITH NATIONAL GUIDELINES
Get PDFStandardizing Oral Anticancer Medication Initiation to Improve Clinician Compliance with National Guidelines (Under the direction of Tracy Vernon-Platt) The aim of this quality improvement project was to standardize the process for initiating oral anticancer medications (OAMs) to align with American Society of Clinical Oncology (ASCO) and Oncology Nursing Society (ONS) guidelines. A quality improvement team was created in accordance with the Iowa Model for Evidence Based Practice. A review of the literature was completed to find effective interventions for initiating OAMs. Next, an anticipatory checklist and standardized EHR documentation tool were created to guide patient education, documentation, and follow up in the initiation of OAMs. These tools were piloted with breast cancer patients of one oncologist at Duke Women’s Cancer Care Raleigh over an 8-week implementation period. By implementing an anticipatory checklist and standardized EHR documentation tool, the QI team improved documentation of ASCO/ONS-mandated topics for 86% of patients initiating an OAM. Comprehensive education using the checklist helped to reduce the patient-initiated phone calls and EHR system messages by 35%. Anticipatory checklists and EHR documentation tools are effective methods for standardizing the process of initiating OAMs. Standardized tools like these should be incorporated into health system policies for OAMs to ensure quality care and patient safety.Doctor of Nursing Practic