As far as the science of organ transplantation has come, rejections do still occur. Overall, rejection happens 10-15% of the time, though this varies by organ. There are various reasons this can occur, and often the trouble takes place before any operation. As good as doctors are, they sometimes select organs that are not right for the particular patient, or not suitable for transplantation at all. With its superior ability to sift through a lot of data, recognize patterns, and make predictions based off that information, AI might be able to help. It could select organs for transplantation much more accurately than humans ever could.
Quality Assessment
The UK’s National Institute for Health and Care Research is putting more than £1 million ($1.22 million) toward the development of an AI-based recognition software called Organ Quality Assessment (OrQA) that uses a database of tens of thousands of images of organs to determine whether donor ones will be suited for transplantation. That’s usually a job for surgeons, but it’s one that a trained AI can do a lot better.
“Currently, when an organ becomes available, it is assessed by a surgical team by sight, which means, occasionally, organs will be deemed not suitable for transplant,” Professor Hassan Ugail, Director of the Centre for Visual Computing at the University of Bradford, said. “We are developing a deep machine learning algorithm which will be trained using thousands of images of human organs to assess images of donor organs more effectively than what the human eye can see.
“This will ultimately mean a surgeon could take a photo of the donated organ, upload it to OrQA and get an immediate answer as to how best to use the donated organ.”
OrQA will then issue donor organs a score based on quality. The NIHR estimates the system could lead to 200 more kidney transplants and 100 more liver transplants annually in the UK. With a transplant list of nearly 7,000 people, that could be a major boon to getting more patients the organs they need.
“Technology has the ability to revolutionize the way we care for people and this cutting edge technology will improve organ transplant services,” Health Minister Neil O’Brien said. “Developed here in the UK, this pioneering new method could save hundreds of lives and ensure the best use of donated organs.”
The National Health Service hopes to conduct a licensing study by 2025 and possibly make OrQA available worldwide.
Donor-Recipient Matching
There are hundreds of classifiers that go into training AI for organ donation, helping the AI get better at processing images, matching donors to recipients, predictive analysis, and real-time immunosuppression. These all contribute to lowering the chances of rejection, ensuring that more organs go to patients in need, and to fulfill their missions of extending lives. Among the things OrQA assesses are evidence of damage or pre-existing conditions and how well the blood has been flushed out of the organs.
In matching donors and recipients, artificial neural networks and random forest AI models have been key. Artificial neural networks mimic the natural architecture of neurons and is good for use with large databases, while random forest models build decision trees and are better for use with small databases.
“Both ANN and RF allow a successful donor allocation in over 80% of D-R pairing, a number much higher than that obtained with the best statistical metrics such as model for end-stage liver disease, balance of risk score, and survival outcomes following liver transplantation scores,” Drs. Javier Briceño, Rafael Calleja, and César Hervás wrote in their paper “Artificial intelligence and liver transplantation: Looking for the best donor-recipient pairing.”
They noted that there are many obstacles to overcome before deep learning-based models match donors and recipients successfully, but that the biggest hurdle is doctors reluctant to give the decision-making power to AI.
University of North Carolina researchers Eric Weimer and Katherine Newhall created the AI-driven Digital Alloimmune Risk Assessment (DARA) to virtually crossmatch donors and recipients, speeding up the process of matching and making it more accurate. They hope to reduce organ waste and add to the more than 40,000 organ transplants that take place in the U.S. annually.
“Overall, this solution streamlines the process by combining several different pieces of information in an easy to read and interpret manner,” Dr. Weimer said. “You can run a high number of patients against a given donor within minutes versus hours. This new process can be done seamlessly, so clinical providers and laboratory professionals all can be notified in part of that decision-making process.”
Benefits & Challenges
Improved matching is just one of the possible benefits AI-based software can bring to organ donation.
“Similar to other clinical decision support systems, AI may help overcome human biases or judgment errors,” Drs. Jeffrey Clement and Angela Q. Maldonado wrote in their paper “Augmenting the Transplant Team with Artificial Intelligence: Toward Meaningful AI Use in Solid Organ Transplant.”
AI could also personalize medication regimens to lower the chances of rejection and assess the risk of patients not upholding their end of the bargain in terms of ensuring successful transplants.
To overcome biases, however, the AI needs better training, since it essentially only knows what people tell it, and bias can unintentionally be programmed into algorithms. It also needs to be more transparent in showing how it arrived at a decision so surgeons and clinicians can understand and explain the reasoning. This in turn will make them more likely to trust input from AI and rely more on it. Doctors can still make the ultimate call, but AI can help ensure it’s the right one.
“To move AI forward, we encourage centers to begin engaging with it, and transplant researchers to include implementation considerations in future AI studies,” Clement and Maldonado wrote.
It could help many more organ recipients live long, healthy lives.
Leave a Reply