AI Predicts Chronic GVHD Risk After Stem Cell Transplant

Chronic GVHD, Stem Cell Transplant, Bone Marrow Transplant, AI in Transplant Medicine, Immune Biomarkers, Transplant Complications, Precision Medicine, Hematology, Oncology Research, Clinical Decision Support, transplant mortality risk, machine learning medicine, GVHD prediction, post-transplant care, precision transplant medicine, hematology oncology, clinical risk models, transplant immunology, BIOPREVENT

Key Takeaways

A new AI-driven tool, BIOPREVENT, predicts chronic GVHD and transplant-related mortality before symptoms appear
Combines immune biomarkers, clinical data, and machine learning
Designed to support early monitoring and research, not treatment decisions
Freely available as a web-based clinical risk assessment tool

Why Chronic GVHD Continues to Challenge Long-Term Stem Cell Transplant Outcomes

Chronic graft-versus-host disease (GVHD) remains one of the most serious long-term complications following stem cell and bone marrow transplantation. While transplantation can be lifesaving, many patients develop chronic GVHD months after discharge, often without early clinical warning signs. This delay limits opportunities for preventive monitoring and timely intervention.

Researchers from MUSC Hollings Cancer Center, in collaboration with national transplant experts, have developed an artificial intelligence, based risk prediction tool that may help clinicians identify patients at high risk for chronic GVHD well before symptoms emerge.

How AI and Immune Biomarkers Are Improving Early Chronic GVHD Risk Prediction

Led by Sophie Paczesny, alongside investigators from the Center for International Blood and Marrow Transplant Research at the Medical College of Wisconsin, the team analyzed data from 1,310 transplant recipients enrolled in four multicenter studies.

The BIOPREVENT model integrates:

Seven immune-related blood biomarkers linked to inflammation, immune activation, and tissue injury
Nine validated clinical factors, including age, transplant type, disease indication, and prior complications

Blood samples were collected 90–100 days post-transplant, a critical period when immune dysregulation may already be underway. Using Bayesian additive regression trees, the model outperformed traditional statistical approaches, particularly in predicting transplant-related mortality.

Clinical Value for Risk Stratification

BIOPREVENT reliably categorized patients into low- and high-risk groups, with outcome differences extending up to 18 months post-transplant. Importantly, the study demonstrated that distinct biomarkers predict chronic GVHD versus transplant-related death, highlighting different biological pathways.

The tool is now available as a free, web-based application, allowing clinicians and researchers to generate individualized risk curves using patient-specific data.

Supporting Precision Transplant Medicine

While BIOPREVENT is not yet intended to guide treatment decisions, it offers a structured framework for risk assessment and clinical trial design. Ongoing studies will determine whether early interventions based on AI-generated risk signals can improve long-term transplant outcomes.