AI-Driven Clinical Decision Support to Reduce Hospital-Acquired Venous Thromboembolism: A Trial Protocol

Hospital-acquired venous thromboembolism (HA-VTE) remains a leading cause of preventable death among hospitalized adults in the US. Despite numerous attempts to prognosticate HA-VTE with risk models, no single model has outperformed the rest, and the effectiveness of such models to drive prophylaxis decisions is unknown. Testing such systems in urban and rural settings may inform their generalizability.

To conduct a randomized clinical trial to assess the effectiveness of artificial intelligence (AI)-driven clinical decision support (CDS) in reducing HA-VTE incidence in adults across urban and rural hospital settings.

This parallel-group, single-blind, pragmatic randomized clinical trial is planned to be conducted from October 1, 2025, through September 30, 2027, by the Vanderbilt University Medical Center, a major academic health system in Tennessee. The study population will include adult (aged ≥18 years) patients admitted to medical, surgical, and intensive care units who may be at high risk for VTE and with no active or contraindication to deep vein thrombosis prophylaxis at Vanderbilt Adult Hospital in urban Nashville and 3 affiliated hospitals serving rural communities in Middle Tennessee.

Patients will be randomized 1:1 within the electronic health record to receive either VTE-AI-driven CDS (nudge practice alert [intervention arm]) or standard care using traditional risk assessment (control arm).

The primary outcome will be incidence of HA-VTE. Secondary trial outcomes will be process metrics, including length of stay, readmission rates, safety, and bleeding events. Outcomes will be analyzed using descriptive statistics and compared using Poisson regression.

Using a validated prognostic model, this study is one of the first to provide insights into whether AI-driven CDS can effectively reduce HA-VTE incidence without increasing adverse events. This study also is intended to provide insights into the usefulness of the same AI model implemented across urban and rural settings. The study's findings and statistical code will be shared with the public through peer-reviewed publication and ClinicalTrials.gov.

ClinicalTrials.gov Identifier: NCT06939803.