The AID-KT project will explore how doctors and patients can use artificial intelligence (AI) to help them to make decisions about organ offers for transplant.

Organ shortage and variations in waiting time

Many patients with kidney disease will benefit from a transplant, but there is a shortage of organ donors meaning average waiting times of nearly 500 days. Waiting time varies across the UK, mainly due to the difficulty deciding whether a kidney offer is suitable for transplant. Offers of kidneys for transplant can come at any time of day or night. The transplant team have limited time and information to decide whether to accept a kidney offer or not.

Transplant decision-making

Doctors currently use their clinical experience to make decisions about organ offers. The NHS has called for tools to be developed to support them in their decision making. Our teams have been developing artificial intelligence (AI) tools to help predict patient outcomes if an offer is accepted and transplanted, or declined to wait for something better. These tools use donor and recipient data, as well as photographs of organs intended for transplant, to help make predictions. These tools have been designed give doctors and patients clear and complete information about why the prediction has been made.

The AID-KT project - taking AI into clinical practice

In this project, we will further develop these tools, concentrating on improving performance and assessing/reducing the risk of bias for patients of different ethnicities, ages and genders. The resulting tool will be made suitable for use in hospitals, combining donor and recipient data and images to predict the success of transplant. We will then conduct a clinical trial in three transplant centres. Each kidney offer received will be placed at random into one of two groups. In the first group the doctors will make decisions using their clinical experience, as they do currently. In the second, the doctor will use their clinical experience plus information from the AI tool. We will compare the two groups to see how use of the tool affects decision making, transplant rates, transplant outcomes and costs. All patient information will be collected at routine clinic visits, so there will be no extra appointments or tests for people taking part.

Patient involvement

We worked with a pre-application patient group, drawn from a diverse range of backgrounds, to develop this project. The patients involved welcomed greater involvement in decision-making and felt that provision of understandable information was key to this. Most of the patients viewed AI in a positive light and felt that predicted information about likely transplant outcomes would be useful to aid decision-making.

We have two patient co-applicants working with us on this project - Judy Marks and Fez Awan. Both have lived experience of kidney transplantation and have worked on research projects in the past. During the project, we will work with Judy and Fez, as well as a wider patient group, to make sure that our research is accessible and patient-centred. Our PPI work will be led by Jess Scaife, an experienced PPI researcher.