There are more candidates on the waitlist for a liver transplant than there are available organs, yet about half the time a match is found with a donor who dies after cardiac arrest following the removal of life support, the transplant must be cancelled. For this type of organ donation, called donation after circulatory death (DCD), the time between the removal of life support and death must not exceed 30–45 minutes, or the surgeons will often reject the liver because of the increased risk of complications to the recipient.
Stanford Medicine researchers have now developed a machine learning (ML)-based model that predicts whether a donor is likely to die within the time frame during which their organs remain viable for transplantation. The model was found to outperform surgeon judgment and reduced by 60% the rate of futile procurements, which occur when transplant preparations have begun but death happens too late.
“By identifying when an organ is likely to be useful before any preparations for surgery have started, this model could make the transplant process more efficient,” said Kazunari Sasaki, MD, clinical professor of abdominal transplantation and senior author on the study. “It also has the potential to allow more candidates who need an organ transplant to receive one.”
Research lead Sasaki, together with first author Rintaro Yanagawa, MD, at Kyoto University, and colleagues, reported on their studies in The Lancet Digital Health, in a paper titled “Development and validation of a machine-learning model to reduce futile procurements in donations after circulatory death in liver transplantation in the USA: a multicentre study.”
For people with end-stage liver disease, which consists of severe and irreversible damage to the organ, the best treatment option is a transplant. The number of people who need a liver exceeds the number of donors, but the gap is starting to narrow due to devices that carry out normothermic machine perfusion, a technique that keeps organs at the ideal temperature and supplied with oxygen while they travel from the donor to the recipient.
These devices have made it possible for organs from donation after circulatory death to be used for transplants. And while most liver donations come from donors who suffered brain death, the number of donations after circulatory death is growing. “The number of liver transplants keeps going up because of donation after circulatory death, and the waitlist is getting smaller,” Sasaki said. “In the future, it might be possible for everyone who needs a liver transplant to get one from a deceased donor.”
A third type of liver transplantation, living donation, involves removing part of a healthy person’s liver to transplant, which is possible because the liver can regenerate. While “it’s a beautiful story,” Sasaki said of living donation, “any major surgery is not without risk to the healthy donor.”
The challenge to donation after circulatory death, however, is time. While the donor is dying, the blood supply to organs throughout the body can vary and, in some cases, stop altogether, leading to liver damage. The liver contains a network of ducts that squeeze out bile, a fluid that helps us digest food, to the gallbladder and intestines. A long period of time between the cessation of life support and the donor’s time of death is associated with malfunctioning ducts, and this can result in serious complications for transplant recipients.
If the donor’s time of death happens more than 30 minutes after blood flow starts to decrease to the body’s organs, the liver might not be useful for transplantation. “The primary cause of non-utilization is an unacceptably long time between the termination of life-sustaining therapy and organ recovery, known as the donor warm ischaemic time,” the authors further explained. “The length of this period is crucial to the quality of DCD grafts, and protracted times increase the risk of biliary complications—particularly non-anastomotic biliary strictures, which are often referred to as the Achilles’ heel of DCD liver transplantation.”
About half of the possible donors die within the first 30 minutes after life support is removed. When death occurs later, between 30 and 60 minutes after life support ends, surgeons use their judgment to determine which donors are the best candidates by considering the donor’s vital signs, blood work, and neurological information, such as the pupil and gag reflex. Still, about half of the transplantations need to be canceled because death occurred too late. Knowing where to allocate resources, such as normothermic machine perfusion devices, can save money and streamline the workload of transplant health care workers, Sasaki commented.
Despite the growing number of DCD donors, models for predicting progression to death within specific timeframes have not been updated in more than 10 years, the team noted. Predicting the timing of death is “complex and multifactorial,” the investigators further pointed out, which limits the capacity of traditional statistical methods.
To predict the time of death, the model uses an array of clinical information from the donor, including gender, age, body mass index, blood pressure, heart rate, respiratory rate, urine output, blood work test results, and cardiovascular health history. The model also considers the ventilator settings, which indicate how much help someone needs to breathe, in addition to neurological assessments of how conscious the patient is, as well as pupil, corneal, cough, gag, and motor reflexes. “We aimed to develop and validate a machine-learning model for predicting progression to death in DCD donors and to evaluate the ability of this model to reduce futile procurements compared with existing statistical models and surgeon predictions.”
To do this, the research team pitted numerous machine-learning algorithms against each other to find the one that best predicted the time of death using the same information available to surgeons. The winning algorithm was more accurate than surgeons and other available computerized tools at predicting whether the donor’s time of death would happen within the acceptable time frame for a successful transplant. The model was trained and validated on more than 2,000 real-world cases from six U.S. transplant centers.
The model, developed using the Light Gradient Boosting Machine (LightGBM) framework, accurately predicts the donor’s time of death 75% of the time, outperforming both existing tools and the average judgment of surgeons, who accurately predicted the time of death 65% of the time. The model also makes accurate predictions for cases with information missing from the medical record.
“The model was validated both retrospectively and prospectively with data from multiple transplant centres, and showed higher accuracy than previous models (the DCD-N score and the Colorado Calculator) and surgeon judgement,” the authors reported. They designed the model to be customizable so it can handle different surgeon preferences and hospital procedures. “The model also effectively manages missing data and allows for center-specific customization of decision thresholds,” the team added.
For example, the model can be set to calculate the time of death from when life support is removed or from when agonal breathing, a gasping breathing pattern that happens as a body is dying, begins. “The researchers have also developed a natural language interface, similar to ChatGPT, that pulls information from the donor medical record into the model. “Our pilot chatbot, as well as aiding data extraction, could also potentially be integrated with the prediction model to automate data input and index generation, substantially reducing manual effort,” the team suggested. In conclusion, the team noted, “The LightGBM model showed potential to reduce futile procurements, reducing their incidence by 60% compared with surgeon predictions and performing effectively even in complex cases.”
Sometimes death unexpectedly occurs within the time frame when organs are suitable for transplantation, but because preparations must be underway before the donor dies, these cases do not result in a transplant. The rate of these missed opportunities was similar for the model and surgeon judgment, at just over 15%.
Because artificial intelligence is rapidly advancing, the researchers expect that the model’s accuracy in predicting time of death will improve and that it will catch more missed opportunities. “We are now working on decreasing the missed opportunity rate because it is in the patients’ best interest that those who need transplants receive them,” Sasaki said. “We continue to refine the model by having competitions among available machine learning algorithms, and we recently found an algorithm that achieves the same accuracy in predicting the time of death but with a missed opportunity rate of about 10%.”
The research team is also working on variations of the model for use in heart and lung transplants.
The post AI Model Could Boost Liver Transplant Efficiency by Predicting Donor Death Timing appeared first on GEN – Genetic Engineering and Biotechnology News.
