A new artificial intelligence tool called BIOPREVENT can predict serious health problems that may arise after stem cell and bone marrow transplants months before symptoms begin. The results of the system, developed by researchers at MUSC Hollings Cancer Center, were published in the Journal of Clinical Investigation. The tool analyzes patient data collected approximately 90 to 100 days after transplantation to provide early warnings about the risk of chronic graft-versus-host disease, which arises from an inadequate immune system, and transplant-related mortality. This aims to enable doctors to identify at-risk patients before problems become apparent.
BIOPREVENT AI tool detects organ transplant risks months in advance
The research team examined data from a total of 1,310 transplant patients from four major studies. Certain immune-related proteins measured in blood tests were combined with clinical information such as patient age, transplant type, and underlying disease. The AI system analyzed the relationships between these data to calculate which patients carried a higher risk in the later stages. Similar to how a credit scoring system performs risk analysis by looking at past spending and payment habits, BIOPREVENT makes a probabilistic inference about the future from past and current health data.
Different AI methods were tested, and the best-performing model was able to predict the risk of developing chronic graft-versus-host disease within one year with significant accuracy. Even higher accuracy values were achieved for transplant-related mortality risk. The researchers also identified which blood values were more decisive for prediction. This means the system not only gives a risk warning but also considers the biological markers behind the risk.
BIOPREVENT was subsequently validated by testing it on an independent group of patients. The research team converted the system into a free web application. Doctors can enter patient information into this platform to receive personalized risk assessments. The tool successfully categorized patients into low and high-risk groups for periods of up to 18 months. This distinction could be particularly helpful in determining which patients require more frequent monitoring.
At this stage, BIOPREVENT is not intended to directly dictate treatment decisions. Researchers emphasize that it needs to be clarified through clinical trials whether acting on early risk warnings truly improves patient outcomes. In other words, the system is positioned not as a decision-maker, but as a digital support tool that provides an additional layer of information to doctors.
0 Comments: