A French team has developed an automated computer assistant that can correct 40% of incorrect diagnoses of kidney transplant rejection. Explanations.
Artificial intelligence to better diagnose transplant rejection. This is the bet of a team of researchers from AP-HP, Paris Cité University and Inserm. It has just developed software that optimizes the diagnosis of transplant rejection. The results of this innovative tool were recently published in the journal Nature Medicine.
Rejection is the leading cause of graft failure after kidney transplantation. The diagnosis of rejection is based on an international so-called Banff classification, developed in 1991. The medical team must analyze and integrate the histological, immunological or even transcriptomic data (study of messenger RNAs) of patients who should guide the diagnosis and their therapeutic management. “This complexity in the diagnosis of rejection, initially necessary to better understand and define its type and severity, has become a daily problem for doctors, faced with situations where it can be difficult to make a correct diagnosis”, we read in the press release issued by the AP-HP.
Doctors and data scientists together
Consequences ? A growing number of incorrect diagnoses made by doctors and resulting in poor patient care.
The team of French researchers has developed a computer system capable of correcting more than 40% of diagnoses. To achieve this, an international consortium of experts in transplantation, nephrology, anatomopathology and epidemiology studied the scientific literature of the last 30 years on the rules of the Banff classification, data scientists and computer developers then developed an algorithm from of this data.
A diagnosis in a few clicks
Developers and programmers were then able to develop software capable, using this algorithm, of interpreting all the medical data of more than 4,000 kidney transplant patients from European and North American centers. “The results are clear, since more than 40% of diagnoses are corrected and requalified by the machine. This tool will make it possible to better treat patients and also to improve clinical trials and the development of immunosuppressive treatments”affirms Pr. Alexandre Loupy at the Center of Expertise for Multi-Organ Transplantation in Paris, the originator of this software.
Very easy to use, the tool makes it possible to obtain a diagnosis in a few clicks. “This computer assistant, which improves the diagnostic performance of rejection phenomena, has been validated by all international transplant societies, and will therefore soon be used by transplant teams around the world to improve patient care. “, writes the AP-HP.
In 2022, 3,376 shortage of available organs, improving the diagnosis of rejection seems essential.