Nova Scotia

How artificial intelligence might change the way kidney transplant patients are treated

A Halifax kidney specialist has teamed up with computer scientists at Dalhousie University to use artificial intelligence to estimate how long a kidney transplant will last.

Dr. Karthik Tennankore wants to predict how long a transplant will last

Dr. Karthik Tennankore says his preliminary work using machine learning is showing 80 per cent accuracy in determining how long a kidney transplant will last. (CBC)

A Halifax kidney specialist has teamed up with computer scientists at Dalhousie University to use artificial intelligence to estimate how long a kidney transplant will last.

Transplants are life-changing for patients, but over time, Dr. Karthik Tennankore says some patients end up back on dialysis when their new kidneys start to fail. His goal is to predict when that will happen.

Tennankore said data analysis called machine learning is showing strong accuracy in predicting success rates, and it could help patients receive better care after the surgery.

"If we can find these groups that are more vulnerable or higher risk, we can better their care," he said.

Tennankore, a nephrologist for Nova Scotia Health, has just finished a study based on a database of 50,000 kidney transplants in the United States. He's now going to apply the same program to patients in Nova Scotia.

There are 180 people in Atlantic Canada on the waiting list for a kidney transplant. Tennankore's research could give them an idea of how long their new organ may last. (Philippe Marcou/AFP/Getty)

He uses characteristics of donors and recipients to figure out connections and make predictions on how long a transplanted kidney will last. Factors such as weight, age and other health conditions play a role in the success of a transplant.

The program also considers how long the kidney was put on ice during the procedures.

"There may be 30, 40, maybe even as high as 50 variables that we combine together and use this machine learning approach to see how they connect all together."

He said the result is the computer can categorize the potential life of the transplant in years. 

Tennankore's goal is to identify higher-risk situations. He said that way doctors can change their treatment programs after the surgery.

That might mean more frequent appointments, or altering their medication.

"What we have shown with our preliminary work, is as high as 80 per cent accuracy, 80 per cent predictability."

Ethical concerns

Tennankore is firm, however, that he does not want the work to play a role in determining who gets a transplant. Some characteristics, such as race, automatically make transplants higher risk.

Tennankore does not want patients to be punished for factors that are out of their control.

"We don't want to deny that patient the opportunity or the chance to get that kidney," he said. "So they get the transplant, but we know that things may not go well in the future, so it's important to us to monitor that patient a bit more closely."

He said he plans to consult with the patient network through the Canadian Donation and Transplantation Research Program to make sure the results are being used in an ethical manner.

Tennankore's team includes Raza Abidi, the leader of the machine learning for the research, and Dr. Amanda Vinson, another Halifax-based nephrologist. They've been given a grant from the Nova Scotia Health Research Fund to study local patients.

Eventually, Tennankore wants to expand his research to patients in Atlantic Canada, all of whom are treated at the transplant centre in Halifax.

He said the team is thrilled to connect the medical world to computer science, ultimately with the hope of helping patients.

"This is absolutely tremendous. This is a new area for me."

ABOUT THE AUTHOR

Carolyn Ray

Videojournalist

Carolyn Ray is a videojournalist who has reported out of three provinces and two territories, and is now based in Halifax. You can reach her at Carolyn.Ray@cbc.ca

Comments

To encourage thoughtful and respectful conversations, first and last names will appear with each submission to CBC/Radio-Canada's online communities (except in children and youth-oriented communities). Pseudonyms will no longer be permitted.

By submitting a comment, you accept that CBC has the right to reproduce and publish that comment in whole or in part, in any manner CBC chooses. Please note that CBC does not endorse the opinions expressed in comments. Comments on this story are moderated according to our Submission Guidelines. Comments are welcome while open. We reserve the right to close comments at any time.

Become a CBC Member

Join the conversationCreate account

Already have an account?

now