How AI Can Help Minimize the Risk of Kidney Transplantation
AI has become a source of hope for many individuals seeking kidney transplants. Traditional methods of evaluating graft failure risks in kidney transplants rely on HLA mismatches. However, a research team from the University of Pennsylvania has developed a new machine-learning algorithm that can identify connections between amino-acid mismatches (AA-MMs) and the likelihood of graft failure.
The algorithm is called FIBRES (Feature Inclusion Bin Evolver for Risk Stratification). It uses evolutionary algorithms to automatically construct AA-MM bins, which helps accurately stratify transplant pairs into high-risk and low-risk groups for graft survival. The researchers tested the algorithm using a dataset of 166,754 kidney transplants and found that it outperformed traditional methods in identifying low-risk patients.
FIBRES optimizes the fitness of AA-MM bins by using an evolutionary algorithm. It selects the best-performing bin as the “parent” and generates new offspring bins by combining and modifying AA positions within the bins. The algorithm also includes a “risk strata minimum” to ensure reliable statistical results.
The researchers conducted three analyses using the FIBRES approach: constructing bins using AA-MMs across five HLA loci, binning AA-MMs within each HLA separately, and evaluating performance using cross-validation. The results showed that FIBRES improved risk stratification compared to the traditional 0-ABDR antigen mismatch method. AA-MM assessment identified 24.4% of kidney transplants as low risk, whereas the 0-ABDR method only identified 9.1% as low risk.
Although FIBRES shows promise in identifying AA-MMs that impact risk, it requires larger datasets for further validation. The researchers plan to address this limitation by expanding the binning to additional HLA loci and comparing results between first transplant and re-transplant recipients. They also aim to optimize bins to stratify donor/recipient pairs into various risk groups and determine the importance of a given MM.
In conclusion, AI has the potential to revolutionize the evaluation and risk assessment of kidney transplants. FIBRES is a promising machine-learning algorithm that can improve the accuracy of graft failure risk stratification. With further development and validation, it could significantly minimize the risk for individuals undergoing kidney transplantation.
To learn more about this research, you can check out the paper and reference article.