Causal pathways and machine learning innovations to improve emergency department outcomes - emergency departments (EDS) are critical for providing immediate care, especially for low-resourced…

Active Grant Awarded

Award ID
R01MD021871
Recipient
Awarding agency
Obligated
$753,676

Award facts

Completion date 5% complete
Place of performance
Pittsburgh, PA 15213, USA

Award details come from USAspending.gov.

About this award

The Department of Health and Human Services National Institutes of Health National Institute for Minority Health and Health Disparities awarded Grant R01MD021871 to the University of Pittsburgh - Of The Commonwealth System of Higher Education (doing business as The University of Pittsburgh) on September 9, 2026, with $753,676 obligated under the Minority Health and Health Disparities Research program (CFDA 93.307).

The award funds research on causal pathways and machine learning to improve emergency department outcomes. Aim 1A uses Bayesian networks to model probabilistic dependencies among demographic, socioeconomic, and clinical factors, drawing on national survey data and UPMC electronic health records to identify causal pathways behind disparities in hospital admission rates, ICU utilization, repeat visits, and imaging and laboratory test use. Aim 1B conducts a matched cohort analysis using 1:1 matching to evaluate differences in waiting times, diagnostic availability, and disposition decisions, with detailed chart reviews where matched patients have differing outcomes. Aim 2 develops and validates a machine learning model to support attainment of recommended care by integrating structured variables, such as vital signs, with clinical notes.

The period of performance runs through June 30, 2030, and work is performed in Pittsburgh, Pennsylvania (ZIP 15213, USA).

Description

Causal pathways and machine learning innovations to improve emergency department outcomes - emergency departments (EDS) are critical for providing immediate care, especially for low-resourced populations. Despite this essential role, persistent disparities in ED outcomes and resource utilization remain, affecting different patient populations. For example, different patient populations often experience higher hospital and ICU admission rates and face differences in the availability of diagnostic and treatment resources. These disparities arise from complex, interconnected factors such as socioeconomic differences, differences in clinical decision-making, and disparities in insurance coverage. Addressing these disparities requires innovative, data-driven approaches that not only identify causal mechanisms but also support consistent and evidence-based decision-making and resource allocation. Aim 1A employs bayesian networks to model probabilistic dependencies among demographic, socioeconomic, and clinical factors, using national survey data and upmc electronic health records (EHRS). This approach will identify causal pathways that contribute to disparities in ED outcomes, such as hospital admission rates, ICU utilization, and repeat visits, as well as differences in resource use like imaging and laboratory tests. Aim 1B involves a matched cohort analysis comparing ED care processes and treatment disparities among study comparison groups. Through 1:1 matching, this analysis will control for confounding factors and evaluate differences in waiting times, diagnostic availability, and disposition decisions. For cases where matched patients have differing outcomes, detailed chart reviews will uncover underlying decision-making patterns and clinical rationale. Aim 2 focuses on developing and validating a machine learning model designed to support attainment of recommended care that integrates structured variables (e.g., vital signs) with clinical notes. This model will employ advanced model development techniques to address differences in model performance arising from existing data, ensuring recommendations that support attainment of recommended care across demographic groups while maintaining predictive accuracy. Aim 3 involves a comparative chart review of predictions from the proposed and standard machine learning models. This will be followed by semi-structured interviews with ED clinicians, administrators, and other collaborators to assess the feasibility, ethical considerations, and practical implications of implementing machine learning-based clinical decision support in real-world ED settings. This project has the potential to improve emergency department decision-making by identifying causal mechanisms underlying disparities and developing machine learning tools to support evidence-based clinical decision-making.

Quick answers

Who received award R01MD021871?

R01MD021871 was awarded to University Of Pittsburgh - Of The Commonwealth System Of Higher Education by National Institute for Minority Health and Health Disparities (National Institutes of Health) on September 9, 2026.

How much is award R01MD021871 worth?

$753,676 has been obligated on R01MD021871.

On GovTribe

See this award's full record

  • Every transaction and subaward
  • The recipient's other awards and teaming partners
  • Similar grant awards
  • Ask GovTribe AI about this award