A powerful artificial intelligence tool could give clinicians a head start in identifying life-threatening complications after stem cell and bone marrow transplants, according to research from MUSC Hollings Cancer Center.
For many patients, a stem cell or bone marrow transplant is lifesaving. But recovery does not end when patients leave the hospital. Serious complications can emerge months later, often without warning.
One of the most challenging is chronic graft-versus-host disease, or GVHD, in which immune cells from the transplant attack a patient’s healthy tissues. The disease can affect multiple organs, causing long-term disability or even death.
Now, researchers, led by Sophie Paczesny, co-leader of the Cancer Biology and Immunology Research Program at Hollings, together with Michael Martens and Brent Logan at the Center for International Blood and Marrow Transplant Research at the Medical College of Wisconsin, have developed an AI-based tool designed to identify patients at higher risk for chronic GVHD before symptoms appear.
The tool—called BIOPREVENT—combines immune biomarkers, clinical data, and machine learning for real-world prediction of a patient’s risk of developing chronic GVHD or dying from transplant-related causes.
A window of opportunity after transplant
Despite advances in transplant care, chronic GVHD remains a leading cause of illness and death following transplant. Yet the biological changes that lead to the disease begin long before symptoms appear.
The first few months after transplant are particularly critical. Patients may feel well, but immune activity beneath the surface can already be setting the stage for complications.
“By the time chronic GVHD is diagnosed, the disease has often been unfolding for months, quietly hurting the body,” Paczesny said. “We wanted to know whether we could detect warning signs earlier, before patients feel sick, giving clinicians more time to intervene.”
Turning blood tests into early warnings
To build the model, the researchers analyzed data from 1,310 stem cell and bone marrow transplant recipients across four large, multicenter studies. Blood samples collected 90 to 100 days after transplant were tested for seven immune proteins linked to inflammation, immune activation, and tissue injury. These biomarkers had been identified and validated in an earlier study led by Paczesny.
The biomarkers were combined with nine clinical factors drawn from transplant registries maintained by the Center for International Blood and Marrow Transplant Research. Transplant centers in the U.S. are required to submit detailed patient data to the registry, providing a standardized source of high-quality clinical information.
The team tested several machine-learning approaches to determine which best predicted patient outcomes. The top-performing model, based on a statistical technique called Bayesian additive regression trees, became the foundation for BIOPREVENT.
From clinical algorithm to real-world tool
Models combining blood biomarkers with clinical data consistently outperformed models based on clinical data alone. The team validated the tool in an independent group of transplant recipients, confirming that it predicted risk beyond the patients used to build the model.
BIOPREVENT also separated patients into low- and high-risk groups, with clear differences in their outcomes up to 18 months later. Notably, different biomarkers predicted different transplant outcomes, underscoring that chronic GVHD and transplant-related death may be driven by distinct biological factors.
To extend the research beyond the study, the team developed BIOPREVENT into a free, web-based application. Clinicians can enter a patient’s biomarker values and clinical data to receive personalized estimates of transplant risk over time.
“It was important that this not remain a theoretical model or a tool limited to a single institution,” Paczesny said. “Making BIOPREVENT freely available helps ensure that researchers and clinicians can test it, learn from it and, ultimately, improve care for transplant patients.”
A step toward more personalized transplant care
For now, BIOPREVENT is intended to support risk assessment and clinical research rather than guide treatment decisions. The next step will be clinical trials to determine whether acting on these early warning signals, such as closer monitoring or preventive therapies for high-risk patients, could improve long-term outcomes.
More broadly, the study reflects a shift toward precision medicine in transplant care, using data to tailor follow-up to each patient’s individual risk.
“This isn’t about replacing clinical judgment,” Paczesny emphasized. “It’s about giving clinicians better information earlier so they can make more informed decisions.”
Although additional validation is needed, the approach represents an important step toward preventing one of transplant medicine’s most serious complications.
“For patients, the uncertainty after transplant can be incredibly stressful,” Paczesny said. “We hope tools like BIOPREVENT can help us see what’s coming sooner and eventually lessen the toll of chronic GVHD.”
Michael J. Martens, Debjani Dutta, Yongzi Yu, Lisa E. Rein, Jerome Ritz, Brent R. Logan and Sophie Paczesny. The BIOPREVENT machine learning algorithm predicts chronic graft-versus-host disease and mortality risk using post-transplant biomarkers. Journal of Clinical Investigation [16 February 2026]. doi: 10.1172/JCI195228.
Grants from the National Cancer Institute (R01CA264921), National Heart, Lung, and Blood Institute (U10HL069294, U24HL138660, P01HL158505) and Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01HD074587) supported this research.
