New AI Model Predicts Donor Viability and Could Cut Wasted Organ Transplant Efforts by 60%

Key Takeaways: 

• A new machine learning model developed at Stanford University predicts whether a donor is likely to die within the critical timeframe needed for safe organ recovery.
  
• The system reduced futile liver procurement attempts by 60% and outperformed senior transplant surgeons.
  
• The tool could improve efficiency, reduce resource waste and expand access for people waiting for a donor organ.
  
  

A data-driven approach to a long-standing challenge

Thousands of people worldwide remain on transplant waiting lists, with demand far exceeding the supply of suitable donor organs. For people who require a liver transplant, recent advances have broadened access by enabling the use of donors who die following cardiac arrest. These cases, known as donations after circulatory death (DCD), have significantly increased potential donor numbers.

However, almost half of DCD liver transplant procedures are cancelled. In every case, timing is critical. After life support is withdrawn, the donor must die within 45 minutes to protect liver viability. If death occurs outside this narrow window, surgeons often reject the organ because of the increased risk of complications for the recipient.

This contributes to substantial resource waste, operational strain on transplant centres and missed opportunities for people waiting for life-saving surgery.

A new predictive tool outperforms top surgeons

Researchers, clinicians and scientists at Stanford University have developed a machine learning model designed to improve prediction accuracy around donor viability. The tool estimates whether a donor is likely to die within the period during which their organs remain suitable for transplantation.

The model surpassed the predictions of highly experienced surgeons and reduced the rate of futile procurements by 60%. Futile procurements occur when surgical teams begin preparing for a transplant but cannot proceed because the donor dies too late for the organ to remain viable.

Dr Kazunari Sasaki, clinical professor of abdominal transplantation and senior author of the study, explained the significance of the advance. “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,” he said. “It also has the potential to allow more candidates who need an organ transplant to receive one.”

The findings were published in The Lancet Digital Health. 

How the model works

The machine learning tool was trained using data from more than 2,000 donors across multiple US transplant centres. It analyses neurological, respiratory and circulatory indicators to estimate a donor’s progression towards death more accurately than previous tools or clinical judgment alone.

During retrospective and prospective testing, the model maintained strong predictive accuracy even when some donor data were missing. Researchers emphasised that this makes it especially practical for real-world clinical settings, where data completeness can vary.

Addressing resource strain and improving outcomes

Currently, transplant centres primarily rely on surgeons’ judgment to assess whether a donor is likely to die within the necessary timeframe. These predictions can vary considerably and may lead to unnecessary preparation of operating theatres, mobilising teams and allocating resources that ultimately go unused.

A reliable, data-driven tool has the potential to improve decision-making, reduce operational burden and ensure that efforts are more closely aligned with the likelihood of a successful transplant.

As the research team noted, the model demonstrates “the potential for advanced AI techniques to optimise organ utilisation from DCD donors”.

Next steps

The team now plans to adapt and test the model for heart and lung transplantation. If successful, this approach could transform prediction processes across multiple organ types, improving access for people waiting for donor organs and enhancing the efficiency of transplant systems worldwide.