INTRODUCTION:Cannabis use is increasing among older adults, but its impact on postoperative pain outcomes remains unclear in this population. We examined the association between cannabis use and postoperative pain levels and opioid doses within 24 hours of surgery. METHODS:We conducted a propensity score-matched retrospective cohort study using electronic health records data of 22 476 older surgical patients with at least 24-hour hospital stays at University of Florida Health between 2018 and 2020. Of the original cohort, 2577 patients were eligible for propensity-score matching (1:3 cannabis user: non-user). Cannabis use status was determined via natural language processing of clinical notes within 60 days of surgery and structured data. The primary outcomes were average Defense and Veterans Pain Rating Scale (DVPRS) score and total oral morphine equivalents (OME) within 24 hours of surgery. RESULTS:504 patients were included (126 cannabis users and 378 non-users). The median (IQR) age was 69 (65-72) years; 295 (58.53%) were male, and 442 (87.70%) were non-Hispanic white. Baseline characteristics were well balanced. Cannabis users had significantly higher average DVPRS scores (median (IQR): 4.68 (2.71-5.96) vs 3.88 (2.33, 5.17); difference=0.80; 95% confidence limit (CL), 0.19 to 1.36; p=0.01) and total OME (median (IQR): 42.50 (15.00-60.00) mg vs 30.00 (7.50-60.00) mg; difference=12.5 mg; 95% CL, 3.80 mg to 21.20 mg; p=0.02) than non-users within 24 hours of surgery. DISCUSSION:This study showed that cannabis use in older adults was associated with increased postoperative pain levels and opioid doses.

BackgroundRacial disparities in COVID-19 incidence and outcomes have been widely reported. Non-Hispanic Black patients endured worse outcomes disproportionately compared with non-Hispanic White patients, but the epidemiological basis for these observations was complex and multifaceted. ObjectiveThis study aimed to elucidate the potential reasons behind the worse outcomes of COVID-19 experienced by non-Hispanic Black patients compared with non-Hispanic White patients and how these variables interact using an explainable machine learning approach. MethodsIn this retrospective cohort study, we examined 28,943 laboratory-confirmed COVID-19 cases from the OneFlorida Research Consortium’s data trust of health care recipients in Florida through April 28, 2021. We assessed the prevalence of pre-existing comorbid conditions, geo-socioeconomic factors, and health outcomes in the structured electronic health records of COVID-19 cases. The primary outcome was a composite of hospitalization, intensive care unit admission, and mortality at index admission. We developed and validated a machine learning model using Extreme Gradient Boosting to evaluate predictors of worse outcomes of COVID-19 and rank them by importance. ResultsCompared to non-Hispanic White patients, non-Hispanic Blacks patients were younger, more likely to be uninsured, had a higher prevalence of emergency department and inpatient visits, and were in regions with higher area deprivation index rankings and pollutant concentrations. Non-Hispanic Black patients had the highest burden of comorbidities and rates of the primary outcome. Age was a key predictor in all models, ranking highest in non-Hispanic White patients. However, for non-Hispanic Black patients, congestive heart failure was a primary predictor. Other variables, such as food environment measures and air pollution indicators, also ranked high. By consolidating comorbidities into the Elixhauser Comorbidity Index, this became the top predictor, providing a comprehensive risk measure. ConclusionsThe study reveals that individual and geo-socioeconomic factors significantly influence the outcomes of COVID-19. It also highlights varying risk profiles among different racial groups. While these findings suggest potential disparities, further causal inference and statistical testing are needed to fully substantiate these observations. Recognizing these relationships is vital for creating effective, tailored interventions that reduce disparities and enhance health outcomes across all racial and socioeconomic groups.

Background: Racial disparities in COVID-19 incidence and outcomes have been widely reported. Non-Hispanic Black patientsendured worse outcomes disproportionately compared with non-Hispanic White patients, but the epidemiological basis for theseobservations was complex and multifaceted.Objective: This study aimed to elucidate the potential reasons behind the worse outcomes of COVID-19 experienced bynon-Hispanic Black patients compared with non-Hispanic White patients and how these variables interact using an explainablemachine learning approach.Methods: In this retrospective cohort study, we examined 28,943 laboratory-confirmed COVID-19 cases from the OneFloridaResearch Consortium's data trust of health care recipients in Florida through April 28, 2021. We assessed the prevalence ofpre-existing comorbid conditions, geo-socioeconomic factors, and health outcomes in the structured electronic health records ofCOVID-19 cases. The primary outcome was a composite of hospitalization, intensive care unit admission, and mortality at indexadmission. We developed and validated a machine learning model using Extreme Gradient Boosting to evaluate predictors ofworse outcomes of COVID-19 and rank them by importance.Results: Compared to non-Hispanic White patients, non-Hispanic Blacks patients were younger, more likely to be uninsured,had a higher prevalence of emergency department and inpatient visits, and were in regions with higher area deprivation indexrankings and pollutant concentrations. Non-Hispanic Black patients had the highest burden of comorbidities and rates of theprimary outcome. Age was a key predictor in all models, ranking highest in non-Hispanic White patients. However, for non-HispanicBlack patients, congestive heart failure was a primary predictor. Other variables, such as food environment measures and airpollution indicators, also ranked high. By consolidating comorbidities into the Elixhauser Comorbidity Index, this became thetop predictor, providing a comprehensive risk measure. Conclusions: The study reveals that individual and geo-socioeconomic factors significantly influence the outcomes of COVID-19.It also highlights varying risk profiles among different racial groups. While these findings suggest potential disparities, furthercausal inference and statistical testing are needed to fully substantiate these observations. Recognizing these relationships is vitalfor creating effective, tailored interventions that reduce disparities and enhance health outcomes across all racial and socioeconomicgroups

Precision prevention embraces personalized prevention but includes broader factors such as social determinants of health to improve cardiovascular health. The quality, quantity, precision, and diversity of data relatable to individuals and communities continue to expand. New analytical methods can be applied to these data to create tools to attribute risk, which may allow a better understanding of cardiovascular health disparities. Interventions using these analytic tools should be evaluated to establish feasibility and efficacy for addressing cardiovascular disease disparities in diverse individuals and communities. Training in these approaches is important to create the next generation of scientists and practitioners in precision prevention. This state-of-the-art review is based on a workshop convened to identify current gaps in knowledge and methods used in precision prevention intervention research, discuss opportunities to expand trials of implementation science to close the health equity gaps, and expand the education and training of a diverse precision prevention workforce.

Background: Cannabis use is associated with higher intravenous anesthetic administration. Similar data regarding inhalational anesthetics are limited. With rising cannabis use prevalence, understanding any potential relationship with inhalational anesthetic dosing is crucial. Average intraoperative isoflurane or sevoflurane minimum alveolar concentration equivalents between older adults with and without cannabis use were compared. Methods: The electronic health records of 22,476 surgical patients 65 yr or older at the University of Florida Health System between 2018 and 2020 were reviewed. The primary exposure was cannabis use within 60 days of surgery, determined via (1) a previously published natural language processing algorithm applied to unstructured notes and (2) structured data, including International Classification of Diseases codes for cannabis use disorders and poisoning by cannabis, laboratory cannabinoids screening results, and RxNorm codes. The primary outcome was the intraoperative time-weighted average of isoflurane or sevoflurane minimum alveolar concentration equivalents at 1-min resolution. No a priori minimally clinically important difference was established. Patients demonstrating cannabis use were matched 4:1 to non-cannabis use controls using a propensity score. Results: Among 5,118 meeting inclusion criteria, 1,340 patients (268 cannabis users and 1,072 nonusers) remained after propensity score matching. The median and interquartile range age was 69 (67 to 73) yr; 872 (65.0%) were male, and 1,143 (85.3%) were non-Hispanic White. The median (interquartile range) anesthesia duration was 175 (118 to 268) min. After matching, all baseline characteristics were well-balanced by exposure. Cannabis users had statistically significantly higher average minimum alveolar concentrations than nonusers (mean +/- SD, 0.58 +/- 0.23 vs. 0.54 +/- 0.22, respectively; mean difference, 0.04; 95% confidence limits, 0.01 to 0.06; P = 0.020). Conclusion: Cannabis use was associated with administering statistically significantly higher inhalational anesthetic minimum alveolar concentration equivalents in older adults, but the clinical significance of this difference is unclear. These data do not support the hypothesis that cannabis users require clinically meaningfully higher inhalational anesthetics doses.

Objective To assess recent temporal trends in guideline-compliant pediatric lipid testing, and to examine the influence of social determinants of health (SDoH) and provider characteristics on the likelihood of testing in youth. Study design In this observational, multiyear cross-sectional study, we calculated lipid testing prevalence by year among 268 627 12-year olds from 2015 through 2019 who were enrolled in Florida Medicaid and eligible for universal lipid screening during age 9 to 11, and 11 437 22-year olds (2017-2019) who were eligible for screening during age 17-21. We compared trends in testing prevalence by SDoH and health risk factors at two recommended ages and modeled the associations between patient characteristics and provider type on lipid testing using generalized estimating equations. Results Testing among 12-year olds remained low between 2015 through 2019 with the highest prevalence in 2015 (8.0%) and lowest in 2017 (6.7%). Screening compliance among 22-year olds was highest in 2017 (21.1%) and fell to 17.8% in 2019. Hispanics and non-Hispanic Blacks in both age groups had about 2%-3% lower testing prevalence than non-Hispanic Whites. Testing in 12-year olds was 12.3% vs 7.7% with and without obesity, and 14.4% vs 7.6% with and without antipsychotic use. Participants who saw providers who were more likely to prescribe lipid testing were more likely to receive testing (OR = 2.3, 95% CI 2.0-2.8, P < .001). Conclusions Although lipid testing prevalence was greatest among high-risk children, overall prevalence of lipid testing in youth remains very low. Provider specialty and choices by individual providers play important roles in improving guideline-compliant pediatric lipid testing.

BACKGROUND:Potential organ donors often exhibit abnormalities on electrocardiograms (ECGs) after brain death, but the physiological and prognostic significance of such abnormalities is unknown. OBJECTIVES:This study sought to characterize the prevalence of ECG abnormalities in a nationwide cohort of potential cardiac donors and their associations with cardiac dysfunction, use for heart transplantation (HT), and recipient outcomes. METHODS:The Donor Heart Study enrolled 4,333 potential cardiac organ donors at 8 organ procurement organizations across the United States from 2015 to 2020. A blinded expert reviewer interpreted all ECGs, which were obtained once hemodynamic stability was achieved after brain death and were repeated 24 ± 6 hours later. ECG findings were summarized, and their associations with other cardiac diagnostic findings, use for HT, and graft survival were assessed using univariable and multivariable regression. RESULTS:Initial ECGs were interpretable for 4,136 potential donors. Overall, 64% of ECGs were deemed clinically abnormal, most commonly as a result of a nonspecific St-T-wave abnormality (39%), T-wave inversion (19%), and/or QTc interval >500 ms (17%). Conduction abnormalities, ectopy, pathologic Q waves, and ST-segment elevations were less common (each present in ≤5% of donors) and resolved on repeat ECGs in most cases. Only pathological Q waves were significant predictors of donor heart nonuse (adjusted OR: 0.39; 95% CI: 0.29-0.53), and none were associated with graft survival at 1 year post-HT. CONCLUSIONS:ECG abnormalities are common in potential heart donors but often resolve on serial testing. Pathologic Q waves are associated with a lower likelihood of use for HT, but they do not portend worse graft survival.

The National Research Mentoring Network (NRMN) Strategic Empowerment Tailored for Health Equity Investigators (SETH) study evaluates the value of adding Developmental Network to Coaching in the career advancement of diverse Early-Stage Investigators (ESIs). Focused NIH-formatted Mock Reviewing Sessions (MRS) prior to the submission of grants can significantly enhance the scientific merits of an ESI’s grant application. We evaluated the most prevalent design, analysis-related factors, and the likelihood of grant submissions and awards associated with going through MRS, using descriptive statistics, Chi-square, and logistic regression methods. A total of 62 out of 234 applications went through the MRS. There were 69.4% that pursued R grants, 22.6% career development (K) awards, and 8.0% other grant mechanisms. Comparing applications that underwent MRS versus those that did not (N = 172), 67.7% vs. 38.4% were submitted for funding (i.e., unadjusted difference of 29.3%; OR = 4.8, 95% CI = (2.4, 9.8), p-value < 0.0001). This indicates that, relative to those who did not undergo MRS, ESIs who did, were 4.8 times as likely to submit an application for funding. Also, ESIs in earlier cohorts (1–2) (a period that coincided with the pre COVID-19 era) as compared to those who were recruited at later cohorts (3–4) (i.e., during the peak of COVID-19 period) were 3.8 times as likely to submit grants (p-value < 0.0001). The most prevalent issues that were identified included insufficient statistical design considerations and plans (75%), conceptual framework (28.3%), specific aims (11.7%), evidence of significance (3.3%), and innovation (3.3%). MRS potentially enhances grant submissions for extramural funding and offers constructive feedback allowing for modifications that enhance the scientific merits of research grants.

Background. Compared with calcineurin inhibitor–based immunosuppression, belatacept (BELA)-based treatment has been associated with better renal function but higher acute rejection rates. This phase 2 study (NCT02137239) compared the antirejection efficacy of BELA plus everolimus (EVL) with tacrolimus (TAC) plus mycophenolate mofetil (MMF), each following lymphocyte-depleting induction and rapid corticosteroid withdrawal. Methods. Patients who were de novo renal transplant recipients seropositive for Epstein-Barr virus were randomized to receive BELA+EVL or TAC+MMF maintenance therapy after rabbit antithymocyte globulin induction and up to 7 d of corticosteroids. The primary endpoint was the rate of biopsy-proven acute rejection at month 6. Results. Because of an unanticipated BELA supply constraint, enrollment was prematurely terminated at 68 patients, of whom 58 were randomized and transplanted (intention-to-treat [ITT] population: n = 26, BELA+EVL; n = 32, TAC+MMF). However, 25 patients received BELA+EVL‚ and 33 received TAC+MMF (modified ITT population). In the ITT population, the 6-mo biopsy-proven acute rejection rates were 7.7% versus 9.4% in the BELA+EVL versus TAC+MMF group. The corresponding 24-mo biopsy-proven acute rejection rates were 19.2% versus 12.5% in the ITT population and 16.0% versus 15.2% in the mITT population; all events were Banff severity grade ≤IIA and similar between groups. One patient in each group experienced graft loss unrelated to acute rejection. The 24-mo mean unadjusted estimated glomerular filtration rates were 71.8 versus 68.7 mL/min/1.73 m 2 in the BELA+EVL versus TAC+MMF groups. Posttransplant lymphoproliferative disorder was reported for 1 patient in each group. No deaths or unexpected adverse events were observed. Conclusions. A steroid-free maintenance regimen of BELA+EVL may be associated with biopsy-proven acute rejection rates comparable to TAC+MMF.

BACKGROUND: Left ventricular dysfunction in potential donors meeting brain death criteria often results in nonuse of donor hearts for transplantation, yet little is known about its incidence or pathophysiology. Resolving these unknowns was a primary aim of the DHS (Donor Heart Study), a multisite prospective cohort study. METHODS: The DHS enrolled potential donors by neurologic determination of death (n=4333) at 8 organ procurement organizations across the United States between February 2015 and May 2020. Data included medications administered, serial diagnostic tests, and transthoracic echocardiograms (TTEs) performed: (1) within 48 hours after brain death was formally diagnosed; and (2) 24±6 hours later if left ventricular (LV) dysfunction was initially present. LV dysfunction was defined as an LV ejection fraction <50% and was considered reversible if LV ejection fraction was >50% on the second TTE. TTEs were also examined for presence of LV regional wall motion abnormalities and their reversibility. We assessed associations between LV dysfunction, donor heart acceptance for transplantation, and recipient 1-year survival. RESULTS: An initial TTE was interpreted for 3794 of the 4333 potential donors by neurologic determination of death. A total of 493 (13%) of these TTEs showed LV dysfunction. Among those donors with an initial TTE, LV dysfunction was associated with younger age, underweight, and higher NT-proBNP (N-terminal pro-B-type natriuretic peptide) and troponin levels. A second TTE was performed within 24±6 hours for a subset of donors (n=224) with initial LV dysfunction; within this subset, 130 (58%) demonstrated reversibility. Sixty percent of donor hearts with normal LV function were accepted for transplant compared with 56% of hearts with reversible LV dysfunction and 24% of hearts with nonreversible LV dysfunction. Donor LV dysfunction, whether reversible or not, was not associated with recipient 1-year survival. CONCLUSIONS: LV dysfunction associated with brain death occurs in many potential heart donors and is sometimes reversible. These findings can inform decisions made during donor evaluation and help guide donor heart acceptance for transplantation.

Objective This study aimed to develop a natural language processing algorithm (NLP) using machine learning (ML) techniques to identify and classify documentation of preoperative cannabis use status. Materials and Methods We developed and applied a keyword search strategy to identify documentation of preoperative cannabis use status in clinical documentation within 60 days of surgery. We manually reviewed matching notes to classify each documentation into 8 different categories based on context, time, and certainty of cannabis use documentation. We applied 2 conventional ML and 3 deep learning models against manual annotation. We externally validated our model using the MIMIC-III dataset. Results The tested classifiers achieved classification results close to human performance with up to 93% and 94% precision and 95% recall of preoperative cannabis use status documentation. External validation showed consistent results with up to 94% precision and recall. Discussion Our NLP model successfully replicated human annotation of preoperative cannabis use documentation, providing a baseline framework for identifying and classifying documentation of cannabis use. We add to NLP methods applied in healthcare for clinical concept extraction and classification, mainly concerning social determinants of health and substance use. Our systematically developed lexicon provides a comprehensive knowledge-based resource covering a wide range of cannabis-related concepts for future NLP applications. Conclusion We demonstrated that documentation of preoperative cannabis use status could be accurately identified using an NLP algorithm. This approach can be employed to identify comparison groups based on cannabis exposure for growing research efforts aiming to guide cannabis-related clinical practices and policies.

Abstract Objective To investigate the associations between domain-specific cognitive change and objectively measured change in physical activity in a sample of community-dwelling older adults. We hypothesized that domain-specific changes in cognition are associated with longitudinal changes in physical activity levels. Methods We used data from 955 participants from The LIFE Study, a multi-center randomized clinical trial. Longitudinal relationships between cognitive domains (i.e., memory, attention, processing speed, executive function) and accelerometer-measured change in daily minutes of physical activity were examined using multivariable linear regression. Cognitive tests included Modified Mini-Mental State Examination (3MSE), Hopkins Verbal Learning Test (HVLT), Digital Symbol Substitution Test (DSST), Task Switching Test, Eriksen Flanker Test and N-Back Test. Physical activity was measured longitudinally using an accelerometer and was categorized to total, light, and moderate to vigorous according to standardized activity count cut points. Results Our results showed an association between cognitive change in the processing speed domain and change in minutes of total (β = 0.47; 95% CI: 0.04, 0.89; p = 0.034) and light (β = 0.42; 95%CI: 0.03, 0.80; p= 0.033) physical activity. Relationships between all other cognitive assessments and physical activity categories were not statistically significant. Conclusions This study demonstrates an association between a decline in processing speed and a decline in the duration of total and light daily physical activity in older adults with low to moderate physical function.

PurposeDespite a scarcity of potential donors for heart transplantation (HT) in the United States (US), a minority are actually accepted for HT. We evaluated donor characteristics associated with heart acceptance in the US and applied modern analytic methods to improve the prediction of heart acceptance.MethodsWe included potential heart donors in the US from 2005 - 2020 (n = 73,948), a recent subset (n = 4,110, spanning 2015 - 2020) of which was enrolled in the Donor Heart Study (DHS). We identified and compared predictors of acceptance among DHS and other donors using logistic regression, incorporating interaction terms in the non-DHS ("nationwide") cohort to characterize time-varying effects. A prediction model was developed using prospectively-collected donor data in the DHS subset, and implemented in the form of a web-based prediction tool.ResultsPredictors of acceptance for HT were similar in the DHS and nationwide cohorts. A random forest model outperformed other prediction algorithms and the inclusion of previously unmeasured predictors (as captured in the DHS) improved model performance (AUC 0.90). In the nationwide cohort, older donor age has become more predictive of non-acceptance over the last 15 years while other factors - including mild cardiac imaging abnormalities, cocaine use, high troponin, hypertension, and Hepatitis C - have become less influential.ConclusionReal-time prediction of donor heart acceptance may improve efficiency during donor management and allocation. The predictors of donor acceptance for HT have changed significantly over time, highlighting the need to continually re-evaluate and update our model. Despite a scarcity of potential donors for heart transplantation (HT) in the United States (US), a minority are actually accepted for HT. We evaluated donor characteristics associated with heart acceptance in the US and applied modern analytic methods to improve the prediction of heart acceptance. We included potential heart donors in the US from 2005 - 2020 (n = 73,948), a recent subset (n = 4,110, spanning 2015 - 2020) of which was enrolled in the Donor Heart Study (DHS). We identified and compared predictors of acceptance among DHS and other donors using logistic regression, incorporating interaction terms in the non-DHS ("nationwide") cohort to characterize time-varying effects. A prediction model was developed using prospectively-collected donor data in the DHS subset, and implemented in the form of a web-based prediction tool. Predictors of acceptance for HT were similar in the DHS and nationwide cohorts. A random forest model outperformed other prediction algorithms and the inclusion of previously unmeasured predictors (as captured in the DHS) improved model performance (AUC 0.90). In the nationwide cohort, older donor age has become more predictive of non-acceptance over the last 15 years while other factors - including mild cardiac imaging abnormalities, cocaine use, high troponin, hypertension, and Hepatitis C - have become less influential. Real-time prediction of donor heart acceptance may improve efficiency during donor management and allocation. The predictors of donor acceptance for HT have changed significantly over time, highlighting the need to continually re-evaluate and update our model.

Precision prevention embraces personalized prevention but includes broader factors such as social determinants of health to improve cardiovascular health. The quality, quantity, precision, and diversity of data relatable to individuals and communities continue to expand. New analytical methods can be applied to these data to create tools to attribute risk, which may allow a better understanding of cardiovascular health disparities. Interventions using these analytic tools should be evaluated to establish feasibility and efficacy for addressing cardiovascular disease disparities in diverse individuals and communities. Training in these approaches is important to create the next generation of scientists and practitioners in precision prevention. This state-of-the-art review is based on a workshop convened to identify current gaps in knowledge and methods used in precision prevention intervention research, discuss opportunities to expand trials of implementation science to close the health equity gaps, and expand the education and training of a diverse precision prevention workforce.

OBJECTIVE:The goal of this study was to address the absence of evidence-based weight-control programs developed for use with Deaf people. METHODS:Community-based participatory research informed the design of the Deaf Weight Wise (DWW) trial and intervention. DWW focuses primarily on healthy lifestyle and weight through change in diet and exercise. The study enrolled 104 Deaf adults aged 40 to 70 years with BMI of 25 to 45 from community settings in Rochester, New York, and randomized participants to immediate intervention (n = 48) or 1-year delayed intervention (n = 56). The delayed intervention serves as a no-intervention comparison until the trial midpoint. The study collected data five times (every 6 months) from baseline to 24 months. All DWW intervention leaders and participants are Deaf people who use American Sign Language (ASL). RESULTS:At 6 months, the difference in mean weight change for the immediate-intervention arm versus the delayed-intervention arm (no intervention yet) was -3.4 kg (multiplicity-adjusted p = 0.0424; 95% CI: -6.1 to -0.8 kg). Most (61.6%) in the immediate arm lost ≥5% of baseline weight versus 18.1% in the no-intervention-yet arm (p < 0.001). Participant engagement indicators include mean attendance of 11/16 sessions (69%), and 92% completed 24-month data collection. CONCLUSION:DWW, a community-engaged, culturally appropriate, and language-accessible behavioral weight loss intervention, was successful with Deaf ASL users.

Background: Enhancement of diversity within the U.S. research workforce is a recognized need and priority at a national level. Existing comprehensive programs, such as the National Research Mentoring Network (NRMN) and Research Centers in Minority Institutions (RCMI), have the dual focus of building institutional research capacity and promoting investigator self-efficacy through mentoring and training. Methods: A qualitative comparative analysis was used to identify the combination of factors that explain the success and failure to submit a grant proposal by investigators underrepresented in biomedical research from the RCMI and non-RCMI institutions. The records of 211 participants enrolled in the NRMN Strategic Empowerment Tailored for Health Equity Investigators (NRMN-SETH) program were reviewed, and data for 79 early-stage, underrepresented faculty investigators from RCMI (n = 23) and non-RCMI (n = 56) institutions were included. Results: Institutional membership (RCMI vs. non-RCMI) was used as a possible predictive factor and emerged as a contributing factor for all of the analyses. Access to local mentors was predictive of a successful grant submission for RCMI investigators, while underrepresented investigators at non-RCMI institutions who succeeded with submitting grants still lacked access to local mentors. Conclusion: Institutional contexts contribute to the grant writing experiences of investigators underrepresented in biomedical research.

OBJECTIVES/GOALS: Analysis and modeling of large, complex clinical data remain challenging despite modern advances in biomedical informatics. We aim to explore the potential of topological data analysis (TDA) to address such challenges in the context of COVID-19 outcomes using electronic health records (EHRs). METHODS/STUDY POPULATION: In this work, we develop TDA approaches to characterize subtypes and predict outcomes in patients with COVID-19 infection. First, data for >70,000 COVID-19 patients were extracted from the OneFlorida EHR database. Next, enhancements to the TDA algorithm Mapper were designed and implemented to adapt the technique to this type of data. Clinical variables, including patient demographics, vital signs, and lab values, were then used as input to conduct a population-level exploratory analysis with an emphasis on identifying phenotypic subtypes at increased risk of adverse outcomes such as major adverse cardiovascular events (MACE), mechanical ventilation, and death. RESULTS/ANTICIPATED RESULTS: Preliminary Mapper experiments have produced visual representations of the COVID-19 patient population that are well-suited to exploratory analysis. Such visualizations facilitate easy identification of phenotypic subnetworks that differ from the general population in terms of baseline variables or clinical outcomes. In this and subsequent work, we aim to fully characterize and quantify differences between these subnetworks to identify factors that may confer increased risk (or protection from) adverse outcomes. We also plan to validate and rigorously compare the efficacy of this TDA-based approach to common alternatives such as clustering, principal component analysis, and machine learning. DISCUSSION/SIGNIFICANCE: This work demonstrates the potential utility of TDA for the characterization of complex biomedical data. Mapper provides a novel means of exploring EHR data, which are otherwise difficult to visualize and can aid in identifying or characterizing patient subtypes in diseases such as COVID-19.