A new study published in the Journal of Biomedical Informatics has developed a deep learning algorithm to accurately predict the risk factors for patients post-liver transplantation. Led by Li and colleagues, the study utilized a large dataset of demographics, clinical, and biomarker data from 160,360 patients who underwent liver transplants between 1970 and 2018. The algorithm, based on a transformer-based multi-task framework, was able to predict five major risk factors that influence post-transplant outcomes: malignancy, diabetes, rejection, infection, and cardiovascular complications.
One of the notable findings of the study was the significance of body mass index (BMI) in determining diabetes outcomes, highlighting its influence on post-transplant health. Additionally, the authors developed a fairness metric based on a disparity index to ensure more equitable distribution of organs across demographic groups while optimizing outcomes. The algorithm prioritized features, including demographic and biomarker data, that highly impacted specific outcomes. Overall, this study represents a step forward in predicting post-liver transplant risk factors and optimizing patient outcomes while promoting fairness in organ allocation.
