Abstract

Background: In the modern era of drug development, drug manufacturers are regularly introducing novel medications that require a wide range of resources that may increase the burden for Veterans seeking care. The VHA must consider how to provide high-quality care in the context of competing resources, the incremental benefit of novel medications over established treatments, and the needs of Veterans. There is a critical need and opportunity to optimize treatment with established medications to maximize their benefit and provide high-quality care considering competing VHA resources and Veterans’ needs. Significance: This challenge is exemplified in the treatment of inflammatory bowel disease (IBD), a costly and debilitating lifelong disease with rising resource needs. IBD represents an ideal proof-of-concept and real-world example on which to conduct research to address: 1. Patient-centered treatment optimization by putting VA Data to Work for Veterans using novel AI/ML methods, 2. Pre-implementation strategies for adoption, and 3. Optimal approaches to healthcare utilization (e.g., access to care) using the VA as a Learning Healthcare System–All of which are VA HSR&D Priorities, updated 3/2023. Innovation & Impact: IBD consists of Crohn’s disease and ulcerative colitis (UC) and affects nearly 3 million Americans and ~80,000 Veterans. This study addresses several key gaps in the care of Veterans with IBD by: Adapting and validating prediction models for optimizing thiopurines; Assessing the usability, acceptability, and feasibility of an Artificial Intelligence (AI)/Machine Learning (ML)-enabled decision-support systems (AEDS) for thiopurine optimization in the VHA; and Evaluating the impact of an AEDS-optimized thiopurine treatment policy on clinical outcomes. Ultimately, an AEDS for thiopurine optimization in Veterans will provide a pragmatic and objective guide for providers to choose to continue, stop, or modify the course of an IBD therapy to produce optimal outcomes while maximizing resources in IBD. Specific Aims: Using learning health system framework, we will: Aim 1: Adapt and validate AI/ML-based models to predict inadequate immunosuppression in Veterans with IBD using real- world data from the VHA EHR (Data-to-Knowledge). Aim 2: Assess the usability, acceptability, and feasibility of AEDS for thiopurine optimization in Veterans with IBD (Knowledge-to-Practice). Aim 3: To evaluate an AEDS- based treatment policy vs. usual care on clinical outcomes (Practice-to-Data). Methodology: In Aim 1, we will retrospectively identify all Veterans with a diagnosis of IBD on thiopurines in the last 10 years from the VHA Corporate Data Warehouse. We will adapt and externally validate our previous model among Veterans with moderate-severe UC on thiopurine monotherapy and internally validate a new model for Veterans with IBD on combination therapy (thiopurines + anti-TNF biologics). We will use AI/ML-based techniques to assess the performance characteristics of these models and leverage statistical techniques to address imprecise measurements in the outcome. In Aim 2, we will elicit behavioral, technological, and systems facilitators and barriers to successful implementation of AEDS via semi-structured interviews with stakeholders (e.g., Veterans, providers, facility leadership, Community Veterans Engagement Board members) using the Unified Theory of User Acceptance. Findings will be fed back to key health system stakeholders. In Aim 3, we will use reinforcement learning in an off-policy evaluation comparing an AEDS-based treatment policy vs. usual care on risk of non-remission and subsequent clinical outcomes (IBD-related steroid use, hospitalizations, surgeries) in Veterans. Next Steps/Implementation: This study is necessary to develop appropriate pragmatic trials that inform IBD treatment guidelines and increase adoption of AEDS-guided thiopurine optimization in the VHA. As a next step, we will apply for an IIR to deploy a prospective pragmatic trial evaluating the effectiveness of AEDS-guided medication optimization vs. traditional approaches. Ultimately, this work will serve as a template for positioning AEDS strategies at the point of care to improve health outcomes and access for Veterans.

Dimensions for VA

Dimensions for VA is a web-based tool available to VA staff that enables detailed searches of published research and research projects.

Learn more about Dimensions for VA.

VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address.     Search Dimensions for this project

DRA: None at this time.

DRE: None at this time.

Keywords: None at this time.

MeSH Terms: None at this time.