Digital phenotyping using personal digital trace data can help identify behavioral patterns and health risk factors, supporting personalized interventions for public health equity. However, tools and frameworks for comprehensive digital trace data collection are lacking. This paper introduces an open-source platform for Screenomics research, enabling in-situ, real-time capture of multimodal digital traces from users’ smartphones as they go about their everyday lives. The hyper-intensive longitudinal digital trace data obtained with the Stanford Screenomics app – including screenshots, application usage logs, interaction histories, and phone sensor readings (e.g., GPS, motion) – empowers researchers to conduct in-depth observations of a wide variety of individual and social behaviors. By design, the data collection process is unobtrusive, minimizing the well-documented biases embedded in individuals’ self-reports of behavior. The platform consists of two native Android front-end applications and a back-end data repository. The Stanford Screenomics Data Collection application offers extensive customization options, allowing researchers to tailor, with minimal additional development effort, the types of data recorded, sampling frequency, how data are transferred from user smartphone to research server, and upload cadence to match their specific study design and requirements. The platform’s Dashboard application supports research coordinators’ real-time monitoring of participants’ data provision, identifying issues in data collection, and automated reactive communications with participants. The platform’s back-end uses a flexible and scalable NoSQL database (Google Cloud Firestore) for secure HIPAA-compliant data storage. Using illustrative 24-hour digital trace data we demonstrate how the platform expands the range of digital phenotyping studies that researchers can now launch.

Variability in board of health (BoH) empowerment can have important consequences for public health as local regulations are developed, considered, and implemented. This research developed and validated a novel empirical measure of BoH empowerment – termed the empowerment index (EI) – by exploring (a) the socio-political factors that are theorized to influence local BoH empowerment, and (b) the positive or negative consequences of health board empowerment on municipality tobacco retail density rate (TDR), a proxy measure of healthier communities across Massachusetts.

A measurement of empowerment was built based on five variables: word count associated with BoH and references to BoH within bylaws or charter; number and method of attainment of BoH seats; and fy22 budget allocated to BoH, with data sourced from a novel quantitative dataset developed from data collection for each of Massachusetts’ 351 municipalities. Linear probability modeling was performed to evaluate EI’s ability to predict the TDR for the 314 municipalities (89%) that reported a TDR, controlling for Republican 2020 presidential election percentage, race, 2022 Census population, and median income.

Municipalities with higher EI scores are associated with a higher probability of having a lower municipality TDR (p=0.0147, 95% CI= -0.4455, -0.0488). Further, municipalities with a greater percentage of white residents are associated with a higher probability of having a lower TDR in their municipality (p=0.0450, 95% CI= -0.0084, -0.0000). This research can assist local boards of health, particularly in vulnerable population communities, in enhancing their empowerment in establishing stronger policies to reduce tobacco retailer density in their community.

Planned Parenthood is committed to providing equitable health care services and programs to its diverse patient population. Analyzing demographic data helps our organization to identify disparities and advance health equity. Traditional race/ethnicity, gender identity, and sexual orientation categories in data collection and reporting often obscure the nuanced realities of marginalized communities, leading to misrepresentation and data gaps that hinder the effectiveness of public health interventions. Expanded and more inclusive demographic categories help us understand the diverse experiences and needs of underserved populations, and tailor care and outreach more effectively.

Demographic granularity enables the identification of inequities that disproportionately affect certain groups; for example, understanding the distinct health challenges within BIPOC communities, or recognizing the unique healthcare needs of gender diverse individuals, allows for tailored interventions that are more effective and equitable.

This presentation draws on interdisciplinary perspectives from public health, reproductive healthcare, and data equity to share best practices for developing comprehensive demographic categories and implementing a roll-up procedure, which aggregates detailed data into broader categories. This ensures that specific identities are preserved while overall trends and patterns remain visible. This dual approach capacitates more accurate and equitable insights.

Through practical examples and data analysis resources this presentation will equip attendees with actionable steps to advocate for and implement more inclusive data practices within their organizations. Ultimately, this presentation aims to catalyze a paradigm shift towards data equity, ensuring all individuals are seen, heard, and represented in the data that shapes our collective future.

ASPR leads the nation's medical and public health preparedness for, response to, and recovery from disasters and public health emergencies. As such, it has a vested interest in understanding how needs of underserved communities are elevated and incorporated into state and local planning for emergencies. To better understand how the needs of underserved communities were being incorporated into preparedness and response planning, ASPR conducted a health equity baseline analysis across several offices, divisions, and branches using a mixed methods approach that incorporated advanced analytics such as natural language processing.

For this study, the evaluation team interviewed individuals across eight ASPR program areas (n=8) who had significant contact with state and local partners to understand how they conceive of equity and how they are beginning to incorporate thinking around equity into their activities. The evaluation team then designed a data analysis approach using natural language processing to analyze hundreds (n=677) of unstructured response plans and after-action reports. Finally, the team interviewed additional individuals across ASPR (n=17) to gather information about the capabilities ASPR developed during COVID-19 to meet the needs of at-risk individuals and underserved communities. Through this study, ASPR gained baseline information on the current state of response and identified promising practices that were taken to plan and care for underserved communities. This study has been critical to ASPR’s subsequent policy actions to further incorporate health equity into all of ASPR’s work. This presentation will focus upon the process and methods ASPR used to conduct this pilot study.

The Tacoma Pierce County Health Department (TPCHD) uses community-based participatory research (CBPR) methods for public health assessment and evaluation. We learned during our Pierce County COVID-19 Health Equity Assessment, that collaborating with trusted community messengers is essential to understanding and addressing inequities. By co-designing culturally appropriate qualitative assessment with diverse communities, we can better identify the health priorities of marginalized and historically excluded communities.

In 2023, TPCHD partnered with the University of Washington Qualitative Research Core to design and implement qualitative data collection training for community leaders and health department staff. Three training sessions were held: two with community members (n= 58) and one with health department staff (n=26). The curriculum included facilitating focus groups and interviews, aiming to help community organizations in fulfill their goals and evaluate programs while advancing health equity.

A pre- post- survey measured participants’ knowledge and skills and general feedback about the course. An additional follow-up-survey was implemented 6-12 months after the training, assessed the application of new skills and identified further training needs. Participants were incentivized with $100 gift cards.

Post survey results indicated improvements in knowledge about qualitive research (i.e. developing interview guides, avoiding bias, member checking) and increase in confidence in their ability to collect qualitative data (i.e. conduct interviews and focus groups). Ninety five percent of those who responded to the follow-up survey reported that they believe the skills learned in the training have benefited the communities they serve. Participants were interested in future trainings to gain qualitative data analysis skills.

Introduction: Online survey research has become increasingly popular among social science researchers due to its scalability, affordability, and the ability to recruit large numbers of participants. It has also been used to reach underrepresented populations in social science research, including racial and ethnic minoritized and LGBTQ individuals. However, the rise of fraudulent activities, specifically survey bots, poses a significant threat to data integrity and to achieving health equity in historically minoritized populations who are often the focus of online studies. Purpose: The purpose of this analysis was to evaluate the implementation of bot detection strategies in an online survey of Latinx sexually minoritized men (SMM). Methods: The online study involved a cross-sectional survey examining the impact of intersecting oppression on HIV prevention outcomes among Latinx SMM living in the U.S. A total of 11 bot detection indicators were used, including an AI-detection software to analyze open-ended responses. These indicators were applied using two approaches to identify bot-generated responses from human responses. Results: Of the 1147 survey entries, 98% were found to be fraudulent and no study participants were verified via email. Among survey entries ≥20% complete (N=837), 68% had an AI-generated open-ended response, 55% were flagged for speed of completion, and 45% began the survey at the same minute as another survey entry. Conclusion: This analysis suggests survey bots are a pervasive threat to online research. Social science researchers should adopt comprehensive bot detection and prevention strategies to protect data integrity and the communities they research.

No shows pose significant challenges to health centers, impacting financial stability, operational efficiency, disrupting service delivery and continuity of patient care. No shows can have the effect of exacerbating healthcare disparities. Addressing no shows is crucial for optimizing resource allocation, improving patient outcomes, and sustaining effective healthcare delivery in underserved areas and for the medically indigent.

This study employed a sequential mixed-methods design examining secondary data of patients’ appointments (N = ~5000), and semi-structured interviews with healthcare providers and allied professionals in an integrated care community health center.

Aggregated analysis shows 57% appointment completions, 23% cancellations, and 20% no shows. However, stratified descriptive analyses provide cues about the complexity of no-shows revealing temporal dynamics and variations across different service domains. The findings offer insights into patients’ service utilization decisions and possible preferences. The healthcare and allied professionals’ interviews suggest the underlying factors for no shows are multifaceted, highlighting system or agency specific factors as well as social determinants of health.

The findings also highlight service gaps and underscore the importance of tailored interventions and technological innovations in reducing no-shows and mitigating their effects across different service domains.

In 2020, as part of an approach to advance health equity, the Centers for Disease Control and Prevention (CDC) National Center for Chronic Disease Prevention and Health Promotion (NCCDPHP) identified 5 social determinants of health (SDOH) domains related to chronic disease for future programmatic work: built environment, community connections to clinical care, tobacco-free policies, social connectedness, and food and nutrition security. Subsequently, NCCDPHP launched an effort to identify and develop a set of measures for monitoring funded programs’ work in these domains.

Measure development involved a literature review, prioritization process, and multisector review and scoring of 59 measures covering all 5 domains. Thirteen multisector community partnerships (MCPs) conducted the review, applying a real-world public health practice lens to assess the relevance and burden of each measure. MCPs’ ratings were analyzed to create summary scores for each measure, and open-ended feedback was synthesized using rapid qualitative analysis.

Feedback from the MCPs raised issues of relevance, burden of data collection, and equity. Reviewers emphasized the need for measures to align with the equity goals of the community, the value of measuring quality and quantity, and the importance of considering MCPs’ scope of practice. Community review informed refinement of NCCDPHP criteria used for selecting and prioritizing measures.

Reflection on what matters to communities may be important when monitoring progress on SDOH program initiatives. Engaging communities in the review of measures may help advance health equity by leveraging community expertise, sharing power with communities, and creating a system of accountability for equitable intervention approaches.

Introduction: Public health practice generates copious documentation, funding progress reports, strategic plans, and community needs assessments. While key data sources for monitoring and evaluation, practitioners rarely have the bandwidth to comb through large volumes of mostly qualitative data to support real-time continuous program improvement. Content analysis systematically analyzes written documents but can be labor intensive.

Methods: Our team sampled four publicly-available, SDOH accelerator plans to explore approaches to expedite content analysis of real-world public health documents. We compared our analysis to GPT-4o abstraction performance on 20 data elements including identifying SDOH focus areas, funder required sections (e.g., purpose, workplan), and descriptions of priority populations and leadership teams between and across the 4 plans. We also compared resources required for each method.

Results: In preliminary results, GPT-4o demonstrated overall abstraction accuracy of 79% (n= 17 errors). Eight errors were major fabrications and GPT-4o missed details for complex abstractions, such as describing the priority population and leadership team. On average, GPT-4o abstraction required fewer hours than study team abstraction, but resource savings diminished when accounting for time for developing prompts and identifying/correcting errors.

Conclusion: AI tools can help save public health professionals’ time on tasks that support continuous program improvement, including content analysis of strategic plans. However, as tools evolve, users should plan for investing time in (1) refining prompts, because the accuracy of abstraction may be dependent on how specific prompts are to the program being used and (2) quality checking, particularly for the abstraction and synthesis of complex details.