Tool uses artificial intelligence and estimates patient's chance of receiving kidney transplant

Researchers from the Hospital das Clínicas of the Faculty of Medicine of Unesp in Botucatu developed a tool that uses artificial intelligence to calculate the chance of a patient receiving a kidney transplant in a given period of time. This is a difficult calculation to make because the kidney transplant queue does not follow a "registration order", but rather the compatibility between donor and recipient, in addition to the patient's health conditions. 

Data from the Brazilian Association of Organ Transplants (ABTO) indicate that in 2020 the waiting list for a kidney transplant in Brazil was 26.862 people, 14.858 of which were in São Paulo. 

ADVERTISING

Keros

The tool was named Keros and the methodology was published in Magazine Plos One. 

To develop the tool, the group led by nephrologist Luís Gustavo Modelli de Andrade, coordinator of the Transplant Program at Hospital das Clínicas and the Data Sciences Laboratory at the Faculty of Medicine of Unesp, collected information from the last 17 years (between the years 2000 and 2017) of patients who were on the transplant queue in the São Paulo State Department of Health database. 

Kidney transplant in Brazil

In total, almost 50 deceased donor records were included.

ADVERTISING

Living donors represent around 20% of total transplants, but they were not included in the database because they follow other criteria for carrying out the procedure.

By crossing available information, the tool can estimate the patient's chance of being transplanted in the State of São Paulo with an accuracy of around 70%.

To do this, the doctor or patient needs to fill out a Keros form with data such as blood type, HLA typing and recipient conditions, among other information.

ADVERTISING

The answer will be the estimated chance of the patient being transplanted within the period, for example: “the probability of a transplant occurring is 28% in 24 months”.

Knowing how long to wait for a transplant is one of the main anxieties for patients.

“The first thing a patient asks us when they find out they will need a kidney transplant is how long they will be waiting in line. And this is a difficult question to answer because it depends on many factors”, said Andrade, responsible for developing the tool.

ADVERTISING

“This is a predictive model, which uses a database of transplants that have already occurred to carry out simulations and try to anticipate an event. But the big problem with predicting the waiting time is that the database fluctuates too much from year to year, especially due to the number of donors and this can interfere with the predicted result”, ponders nephrologist Lúcio Requião Moura, from the Human Transplant Program. Kidney at Hospital Israelita Albert Einstein and professor at the Federal University of São Paulo (Unifesp).

Source: Einstein Agency

Scroll up