Beyond the Checkbox: Comparing Notes between Academic and Community Clinician Practices for SDoH Capture

Presentation Type: Paper - Student

Click to View Presentation

2026 Symposium 25x5 Presentation

Presentation Time: 03:30 PM - 03:42 PM

Abstract Keywords: Qualitative Methods, Workflow, Documentation Burden
Programmatic Theme: Clinical Informatics

Health systems throughout the United States use a variety of methods to collect and record social drivers of health (SDoH) ranging from paper-based surveys to electronic methods, such as electronic forms, as structured questionnaires, and natural language processing. The collection and recording of SDoH also depend on the settings and priorities of the health system in which they are collected. For example, community health centers (CHC) see a disproportionate share of individuals with greater health disparities compared to ambulatory clinics affiliated with academic medical centers (AMC). Through semi-structured interviews of 13 primary care providers at four ambulatory clinics spanning AMC and CHC settings, five themes emerged: patient provider relationship, conversational dialogue, problem solving with social work referral, fluctuating circumstances, and workflow differences. Three themes were consistent between settings, while two themes differed between health systems. Specifically, differences in how AMC and CHC clinics manage fluctuating circumstances led to workflow differences.

Speaker(s):
Carolin Spice, MS
University of Washington
Author(s):
John Gennari, PhD - University of Washington, Dept of Biomedical Informatics & Medical Education; Andrea Hartzler, PhD FACMI - University of Washington; Patrick Wedgeworth, MD, MISM - University of Washington; Oliver Bear Don't Walk, PhD - University of Washington;

---

Data-Driven Approaches for Improving Health-Related Social Needs Screening

Presentation Type: Podium Abstract

Click to View Presentation

Presentation Time: 03:42 PM - 03:54 PM

Abstract Keywords: Patient-/Person-Generated Health Data, Workflow, Fairness and elimination of bias
Programmatic Theme: Clinical Informatics

We present methods for evaluating health-related social needs screening procedures for healthcare delivery systems. We analyzed EHR data from a large, adult patient sample to assess demographic biases in screening as well as redundancy in the risk information captured across screening items. The findings from these analyses will be used to guide improvements in the quality and efficiency of these screening procedures.

Speaker(s):
John Powers, PhD
University of North Carolina at Chapel Hill
Author(s):
Amy Shaheen, MD, MSc - UNC Health Alliance; Barbara Entwisle, PhD - University of North Carolina at Chapel Hill; Emily Pfaff, PhD, MS - UNC Chapel Hill School of Medicine;

---

Quantifying the Observability of Intervention-Oriented Healthcare Access Barriers in Structured Electronic Medical Records

Presentation Type: Paper - Student

Click to View Presentation

Presentation Time: 03:54 PM - 04:06 PM

Abstract Keywords: Population Health, Diversity, Equity, Inclusion, and Accessibility, Surveys and Needs Analysis, Machine Learning
Programmatic Theme: Clinical Research Informatics

Healthcare access barriers shape health equity and care delivery, yet many intervention-relevant barriers are not routinely captured in structured electronic medical records (EMR). This creates a challenge for health systems: determining which barriers can be inferred from routine clinical data and which require direct patient-reported assessment. Using linked Healthcare Access and Utilization survey responses and OMOP-standardized EMR data from the NIH All of Us Research Program, we operationalized seven access barrier domains and evaluated their predictability from a one-year pre-survey EMR window. We compared direct clinical concept features, interpretable proxy summaries of care patterns and socioeconomic context, and a combined representation using standardized machine learning pipelines. Results revealed a clear gradient in EMR observability: financial and logistical barriers showed measurable signal in clinical records, whereas navigation and health literacy barriers remained largely unobservable. These findings support hybrid measurement strategies combining EMR-based screening with targeted patient-reported assessment.

Speaker(s):
Shuwei He, Master of Science
Carnegie Mellon University
Author(s):
Shuwei He, Master of Science - Carnegie Mellon University; Holly Wiberg, PhD in Operations Research - Carnegie Mellon University; Rema Padman, PhD - Carnegie Mellon University;

---

Community-Based Organizations’ Experiences and Desires of Using Electronic Tools to Facilitate Referrals for Health-related Social Needs

Presentation Type: Podium Abstract
2026 Symposium Health Equity Presentation

Presentation Time: 04:06 PM - 04:18 PM

Abstract Keywords: Health Equity, Qualitative Methods, Surveys and Needs Analysis
Programmatic Theme: Clinical Informatics

Community-based organizations (CBOs) are vital for addressing health-related social needs, yet often rely on fragmented, outdated referral and data-sharing methods. Through interviews with CBOs, this study reveals barriers including limited interoperability, cumbersome electronic tools, lack of access to critical client information, and administrative burden. CBOs desire user-friendly, integrated platforms with two-way communication and comprehensive client data. Findings underscore the need for inclusive design to enhance referral workflows and social care coordination.

Speaker(s):
Bradley Iott, MPH, PhD
University of Michigan
Author(s):
Nancy Yang, MPH - UCSF; Matthew Pantell, MD, MS - University of California, San Francisco;

---

Developing a Precision Feedback System to Support Social and Pediatric Primary Care Integration

Presentation Type: Podium Abstract

Click to View Presentation

2026 Symposium Health Equity Presentation

Presentation Time: 04:18 PM - 04:30 PM

Abstract Keywords: Health Equity, User-centered Design Methods, Qualitative Methods, Healthcare Quality, Patient-/Person-Generated Health Data
Programmatic Theme: Clinical Informatics

In a human-centered design process, we iteratively developed a precision feedback system for pediatric primary care teams to improve health-related social needs screening and referral, across 3 phases: contextual inquiry, iterative prototyping and usability testing, and technical feasibility assessment. Our precision feedback messages were perceived to be acceptable, usable, and useful by prospective users. Technical feasibility was demonstrated by a prototype precision feedback system that generated quarterly messages for 177 pediatric clinics, using real-world performance data.

Speaker(s):
Yidan Cao, MS, MPP
University of Michigan Medical School
Author(s):
Yidan Cao, MS, MPP - University of Michigan Medical School; Farid Seifi, PhD - University of Michigan Medical School; Zach Landis-Lewis, PhD,MLIS - University of Michigan;

---

Analyzing Mobility Functional Status in Health Records with Natural Language Processing: Use Cases in Social Security Administration Disability Benefits Determination

Presentation Type: Paper - Regular
2026 Symposium Health Equity Presentation

Presentation Time: 04:30 PM - 04:42 PM

Abstract Keywords: Disability, Accessibility, and Human Function, Natural Language Processing, Informatics Implementation, Real-World Evidence Generation, Administrative Systems, Information Retrieval, Artificial Intelligence
Programmatic Theme: Clinical Informatics

Locating sparse information about functional status in high volumes of medical text is a key challenge in the US Social Security Administration's process of determining disability benefits. Natural language processing (NLP) methods can help automate discovery of functional status information at scale, but face two significant challenges: the lack of NLP models for analysing semantically and syntactically complex functional status data, and limited compute capabilities in SSA’s secure environment. In this study, we leveraged a flexible, low-resource NLP method optimized for functional status information to retrieve information on mobility limitations, one of the most frequent allegations in disability applications. We achieved over 80% recall on mobility information, and up to 89% rank correlation when ranking documents by mobility density. Experiments on a dataset of long, heterogeneous documents show our approach performs nearly perfectly at surfacing documents with mobility information, indicating strong potential utility for semi-automated triage for functional status information.

Speaker(s):
Denis Newman-Griffis, PhD
University of Sheffield
Author(s):
Denis Newman-Griffis, PhD - University of Sheffield; Jonathan Camacho Maldonado, MD - National Institutes of Health; Pei-Shu Ho, PhD - National Institutes of Health;

---