Faculty Bibliography

We study regularized estimation in high-dimensional longitudinal classification problems, using the lasso and fused lasso regularizers. The constructed coefficient estimates are piecewise constant across the time dimension in the longitudinal problem, with adaptively selected change points (break points). We present an efficient algorithm for computing such estimates, based on proximal gradient descent. We apply our proposed technique to a longitudinal data set on Alzheimer's disease from the Cardiovascular Health Study Cognition Study. Using data analysis and a simulation study, we motivate and demonstrate several practical considerations such as the selection of tuning parameters and the assessment of model stability. While race, gender, vascular and heart disease, lack of caregivers, and deterioration of learning and memory are all important predictors of dementia, we also find that these risk factors become more relevant in the later stages of life.

Latent variable network models that accommodate edge correlations implicitly, by assuming an underlying latent factor, are increasing in popularity. Although, these models are examples of what is a growing body of research, much of the research is focused on proposing new models or extending others. There has been very little work on unifying the models in a single framework. In this paper, we present a complete framework that organizes existing latent variable network models within an integrative generalized additive model. Our framework is called Conditionally Independent Dyad (CID) models, and includes existing network models that assume dyad (or edge) independence conditional on latent variables and other components in the model. We further discuss practical aspects of model fitting such as posterior parameter estimation via MCMC, identifiability of parameters, approaches to handle missing data and model selection via cross-validation, for the proposed additive CID models. Finally, by presenting several data examples, we illustrate the utility of the proposed framework and provide advice on selecting components for building new CID models.

Social network data represent interactions and relationships among groups of individuals. One aspect of social interaction is social influence, the idea that beliefs or behaviors change as a result of one's social network. The purpose of this article is to introduce a new model for social influence, the latent space model for influence, which employs latent space positions so that individuals are affected most by those who are "closest" to them in the latent space. We describe this model along with some of the contexts in which it can be used and explore the operating characteristics using a series of simulation studies. We conclude with an example of teacher advice-seeking networks to show that changes in beliefs about teaching mathematics may be attributed to network influence.

OBJECTIVE:To compare anonymized and non-anonymized approaches for imputing race and ethnicity in descriptive studies of chronic disease burden using electronic health record (EHR)-based datasets. STUDY SETTING AND DESIGN/METHODS:In this New York City-based study, we first conducted simulation analyses under different missing data mechanisms to assess the performance of Bayesian Improved Surname Geocoding (BISG), single imputation using neighborhood majority information, random forest imputation, and multiple imputation with chained equations (MICE). Imputation performance was measured using sensitivity, precision, and overall accuracy; agreement with self-reported race and ethnicity was measured with Cohen's kappa (κ). We then applied these methods to impute race and ethnicity in two EHR-based data sources and compared chronic disease burden (95% CIs) by race and ethnicity across imputation approaches. DATA SOURCES AND ANALYTIC SAMPLE/UNASSIGNED:Our data sources included EHR data from NYU Langone Health and the INSIGHT Clinical Research Network from 3/6/2016 to 3/7/2020 extracted for a parent study on older adults in NYC with multiple chronic conditions. PRINCIPAL FINDINGS/RESULTS: = 0.33). When these methods were applied to the NYU and INSIGHT cohorts, however, racial and ethnic distributions and chronic disease burden were consistent across all imputation methods. Slight improvements in the precision of estimates were observed under all imputation approaches compared to a complete case analysis. CONCLUSIONS:BISG imputation may provide a more accurate racial and ethnic classification than single or multiple imputation using anonymized covariates, particularly if the missing data mechanism is MNAR. Descriptive studies of disease burden may not be sensitive to methods for imputing missing data.

OBJECTIVE:This study examined associations between access to leisure-time physical activity (LTPA) facilities and parks and repeated measures of glycated hemoglobin (A1C) over time, using follow-up tests among United States Veterans with newly diagnosed type-2 diabetes (T2D). METHODS:Data were analyzed from 274,463 patients in the Veterans Administration Diabetes Risk cohort who were newly diagnosed with T2D between 2008 and 2018 and followed through 2023. Generalized estimating equations with a logit link function and binomial logistic regression were used to examine associations. RESULTS:Patients were on average 60.5 years of age, predominantly male (95.0 %) and white (66.9 %), and had an average of 11.7 A1C tests during the study follow-up period. In high- and low-density urban communities, a one-unit higher LTPA facility density score was associated with 1 % and 3 % greater likelihood of in-range A1C tests during follow-up, respectively, but no association was observed among patients living in suburban/small town and rural communities. Across community types, closer park distance was not associated with subsequent greater odds of in-range A1C tests. Unexpectedly, in low-density urban areas, the likelihood of in-range A1C tests was 1 % lower at farther park distances. CONCLUSIONS:These results suggest that broader access to LTPA facilities, but not park proximity, may contribute in small ways to maintaining glycemic control after T2D diagnosis in urban communities. Tailored interventions may be needed to promote patients' effective use of LTPA facilities and parks.

BACKGROUND:Results from national surveys indicate that many older adults reported delayed medical care during the acute phase of the COVID-19 pandemic, yet few studies have used objective data to characterize healthcare utilization among vulnerable older adults in that period. In this study, we characterized healthcare utilization during the acute pandemic phase (March 7-October 6, 2020) and examined risk factors for total disruption of care among older adults with multiple chronic conditions (MCC) in New York City. METHODS:This retrospective cohort study used electronic health record data from NYC patients aged ≥ 50 years with a diagnosis of either hypertension or diabetes and at least one other chronic condition seen within six months prior to pandemic onset and after the acute pandemic period at one of several major academic medical centers contributing to the NYC INSIGHT clinical research network (n=276,383). We characterized patients by baseline (pre-pandemic) health status using cutoffs of systolic blood pressure (SBP) < 140mmHg and hemoglobin A1C (HbA1c) < 8.0% as: controlled (below both cutoffs), moderately uncontrolled (below one), or poorly controlled (above both, SBP > 160, HbA1C > 9.0%). Patients were then assessed for total disruption versus some care during shutdown using recommended care schedules per baseline health status. We identified independent predictors for total disruption using logistic regression, including age, sex, race/ethnicity, baseline health status, neighborhood poverty, COVID infection, number of chronic conditions, and quartile of prior healthcare visits. RESULTS:Among patients, 52.9% were categorized as controlled at baseline, 31.4% moderately uncontrolled, and 15.7% poorly controlled. Patients with poor baseline control were more likely to be older, female, non-white and from higher poverty neighborhoods than controlled patients (P < 0.001). Having fewer pre-pandemic healthcare visits was associated with total disruption during the acute pandemic period (adjusted odds ratio [aOR], 8.61, 95% Confidence Interval [CI], 8.30-8.93, comparing lowest to highest quartile). Other predictors of total disruption included self-reported Asian race, and older age. CONCLUSIONS:This study identified patient groups at elevated risk for care disruption. Targeted outreach strategies during crises using prior healthcare utilization patterns and disease management measures from disease registries may improve care continuity.

OBJECTIVES/OBJECTIVE:Despite significant therapeutic advances in psoriatic arthritis (PsA), many patients do not achieve remission and cycle through multiple biologic (b)- or targeted synthetic (ts)- DMARDs. Identifying the underlying reasons for repetitive therapeutic failure remains a knowledge gap. Here we describe prescribing patterns and characteristics of PsA patients with multi-b/tsDMARD failure at the NYU Psoriatic Arthritis Center. METHODS:Nine hundred sixty PsA patients were enrolled in an observational, longitudinal registry. Demographics, medical history, medication use, and psoriatic disease phenotype were collected. Multi-b/tsDMARD failure was defined as requiring ≥ 4 b/tsDMARDs. RESULTS:Seven hundred twenty-five patients (75%) used ≥ 1 b/tsDMARD during their disease course. The initial b/tsDMARDs prescribed were predominately anti-TNF agents. 166 (17%) patients had multi-b/tsDMARD failure. Compared to those requiring 1 b/tsDMARD, female sex (OR 2.3; 95%CI 1.4-3.8), axial disease (OR 2.1; 95% CI 1.2-3.6), depression (OR 2.0; 95%CI 1.1-3.7), and obesity (OR 1.7; 95%CI 1.0-2.8) were risk factors for multi-b/tsDMARD failure disease after adjustment for age, disease duration, sex, depression, smoking, obesity, and skin severity. Patients with multi-b/tsDMARD failure PsA also had increased disease activity at their clinical visit (i.e., swollen joint count, p = 0.005). CONCLUSION/CONCLUSIONS:In this cohort, 17% patients with PsA experienced multi-b/tsDMARD failure. These patients were more likely to be female, obese, and have higher rates of axial involvement and depression, along with higher active disease activity. This highlights the inflammatory and non-inflammatory drivers of multiple therapeutic failures, underscoring the need for precision medicine strategies and potential non-pharmacologic adjuvant therapies for patients with PsA to improve outcomes and quality of life.

Mandatory prescription drug monitoring programs and cannabis legalization have been hypothesized to reduce overdose deaths. We examined associations between prescription monitoring programs with access mandates ("must-query PDMPs"), legalization of medical and recreational cannabis supply, and opioid overdose deaths in United States counties in 2013-2020. Using data on overdose deaths from the National Vital Statistics System, we fit Bayesian spatiotemporal models to estimate risk differences and 95% credible intervals (CrI) in county-level opioid overdose deaths associated with enactment of these state policies. Must-query PDMPs were independently associated with on average 0.8 (95% CrI: 0.5, 1.0) additional opioid-involved overdose deaths per 100,000 person-years. Legal cannabis supply was not independently associated with opioid overdose deaths in this time period. Must-query PDMPs enacted in the presence of legal (medical or recreational) cannabis supply were associated with 0.7 (95% CrI: 0.4, 0.9) more opioid-involved deaths, relative to must-query PDMPs without any legal cannabis supply. In a time when overdoses are driven mostly by non-prescribed opioids, stricter opioid prescribing policies and more expansive cannabis legalization were not associated with reduced overdose death rates.