Select Page

Predictive analytics project to launch education in December 2022, is intended to increase organ utilization

Predictive analytics project to launch education in December 2022, is intended to increase organ utilization

United Network for Organ Sharing (UNOS) is preparing to incorporate predictive analytics into the organ offer experience for adult deceased donor kidneys. The decision support tool is intended to reduce the number of deceased donor kidneys that go unused and to increase transplants within the current kidney allocation system.

A national rollout starting at the end of 2022 will enable all adult kidney transplant programs to evaluate organ offers through predictive analytic data. The data will display for adult deceased donor kidney offers in DonorNet Mobile℠, a component of UNet℠, the engine that powers the donation and transplantation system.

UNOS is collaborating with Accenture Federal Services on the project, which combines behavioral science, data science and innovative technology to help get the right organ to the right patient at the right time. In 2019, the Organ Procurement and Transplantation Network (OPTN) Ad Hoc Systems Performance Committee recommended investigating predictive analytics at the time of organ offer.

How it works: Using predictive data to improve offer acceptance

Predictive analytics offer the transplant community the ability to customize decision-support tools based on program-specific and patient-specific criteria.

When transplant programs receive organ offers for their kidney patients, each offer must be evaluated and either accepted or declined in DonorNet Mobile. Some transplant teams may decline kidney offers based on their estimation of how quickly their patient is likely to receive a similar or better organ offer.

Data scientists from UNOS and Accenture produced statistical models to predict:

  • Time-to-next-offer: The length of time the candidate is likely to wait while actively registered in the organ matching system until they are primary for another offer of an organ of the same quality as the one being evaluated.
    • Users can choose to view predicted times for a future offer of a kidney with a kidney donor profile index (KDPI) below 50 or a KDPI below 30.
    • The time-to-next-offer model includes additional factors like candidate blood type, candidate sensitivity, or CPRA, time on dialysis, and the candidate’s sequence number on the specific match run. The model also adjusts for geographic differences in offer patterns.
  • Likelihood of death: The probability of the transplant candidate’s death before the predicted time to next offer, if they do not receive the organ for transplant.
    • Users will see a visualization of the candidate’s predicted survival likelihood over the next three years without a transplant.
    • The predicted mortality model includes factors such as the candidate’s age, time on dialysis, and diabetes status

Understanding the potential likelihood of a patient’s death if an organ offer is declined may lead to increased organ acceptance and usage. These predictions are intended to supplement, not replace, existing data and clinical judgment.

National rollout builds upon previous phases

Following a formal presentation to the OPTN Board of Directors at their meeting on Dec. 5, 2022, the education and training on this enhancement will launch to the transplant community, so that all adult kidney transplant programs may begin to fully leverage the modification starting in early January of 2023. The Network Operations Oversight Committee, the Data Advisory Committee and the Kidney Transplantation Committee will also hear presentations on the project before the board’s December meeting.

The national launch of predictive analytics is the fourth phase of a project that has involved a variety of stakeholders in the organ donation and transplantation community, as well as industry-leading information technology consultants.

The first phase took place in the summer of 2021 and involved concept testing, planning and development, including a behavioral study that included surgeons, nephrologists and transplant administrators. A second phase in the fall involved designing and testing the interface.  A third phase in early 2022 implemented predictive analytics to a pilot group of 15 adult kidney programs, in order to gather information and make improvements.

Analyses of overall DonorNet Mobile kidney offer acceptance rates before and during the pilot found no unintended effects on acceptance rates, initial response times, or final response times among the participating transplant programs. Findings from the pilot phase have informed the national rollout, which will launch in December with the release of education and training resources. Predictive analytic models will be developed for other organ types in the future.

Advisory panel will monitor safety

To support the project and provide community input, UNOS formed an advisory panel in November 2021 consisting of clinical experts. The panel regularly receives data monitoring reports and advises on the project, including whether the initiative would ever need to be paused due to unintended consequences.

Patient safety is at the forefront of UNOS research and innovation, and the advisory panel will continue to monitor this project following implementation and provide regular updates to the Board and committee stakeholders.

Improving system performance through innovation

The predictive analytics project supports the strategic goal to improve waitlisted patient, living donor and transplant recipient outcomes. It is one of a number of innovative technology initiatives aimed at increasing transplant through improved system performance. Another project launched nationally this year, Offer Filters, aims to increase the number of transplants by improving placement efficiency, another strategic goal. A driving force of UNOS is to improve the U.S. system so that more lifesaving organs are available for transplant.


For questions about predictive analytics or other research and technology projects, please [email protected]


Share This