One of the first experiments in the world to apply artificial intelligence to the diagnosis of breast and breast cancer and some types of cancer that can be treated with immunotherapy starts in Italy. A way to quickly identify the pathology and choose the most appropriate therapy. The project was started at the University of Padua, in collaboration with a Helsinki company. Aiforia Technologies. The Veneto Region is financing the initiative with 21 million euros. To announce it, on the occasion of World Health Forum Venetoinaugurated yesterday in Padua, the governor Luca Zaia and president of the University School of Medicine, professor Angelo Paolo Dei Tos.
A project that constitutes a concrete novelty on the international scene. “Everyone talks about artificial intelligence and in some ways we are surrounded by it – explains the professor Angelo Dei Tospresident of the School of Medicine of the University of Padua – the difference is that we are the only Pathological Anatomy that has obtained Pnrr funding, giving rise to a massive digitalization process that has no equal in Europe: the Veneto Region will therefore be a pioneer on this front.”
Artificial intelligence, here’s what can be done in healthcare by Gianluigi Grieco, Piero Poccianti 15 March 2024
Man and machine
At the basis of this digital revolution are a series of algorithms that are trained and controlled by researchers. Because the machine can work, but always if governed by real doctors. Artificial Intelligence is ‘trained’ by white coats to recognize pathologies and first associate a diagnosis, then a therapy. “It is the man who teaches the machine what to do and provides it with the contents and data. For this reason it will always be necessary to have trained doctors, with high-level skills, to in turn make the machine ‘special’. We are talking about an application to oncological diagnostics which offers more effective information in the therapeutic field”, adds the president of the School of Medicine of the University of Padua.
The applications of AI
Already today, artificial intelligence has various applications in medicine, because it manages to process different forms of data before arriving at a diagnostic answer. This happens, for example, with x-rays, magnetic resonance imaging and with the use of algorithms to potentially support decision-making processes. Among the many applications is the use of very large databases which can reduce the margins of uncertainty during the choice of therapy.
But the University of Padua project is a step forward. “We have started a massive digitalisation of histological preparations which by the end of the year will be acquired, scanned and online, i.e. usable by all the Pathological Anatomies of the Veneto. This process will help us in making evaluation processes more accurate, for example of biomarkers, in order to indicate the best therapies in oncology – adds Dei Tos – These are processes that are already carried out by pathologists today, but thanks to Artificial Intelligence we will be able to quantify proteins with an accuracy, precision and speed that would otherwise be unthinkable. It will allow a cell count, an evaluation of colors and differences in tissues that the human eye is unable to see.”
The medicine of the future, between artificial intelligence and space exploration 14 March 2024
It must be said that this data processing by Artificial Intelligence will not replace the pathologist. The final evaluation will always be up to him but, thanks to this technology, he will be able to guarantee calculations that will improve accuracy. And all this will make the time needed to decide on the treatment shorter, giving the specialist more space to dedicate to dialogue with the patient.
I 5 algorithms
Three algorithms tested by the University of Padua concern breast cancer and evaluate the expression of estrogen and progesterone, predictive parameters of whether or not a patient is a candidate for hormone therapy. Another algorithm helps to more precisely photograph the level of severity of prostate cancer.
“Many therapeutic decisions are based on the morphological characteristics of the neoplastic mass, they reveal its degree of aggressiveness – explained Dei Tos -. The dedicated algorithm indicates this, allowing the doctor to choose which therapy to adopt or even not to resort to any, but to opt for a conservative solution”.
The wearable AI that restores the voice in case of vocal cord problems by Simone Valesini 21 March 2024
Immunotherapy in the treatment of tumors
Finally, the fifth algorithm evaluates the PDL1 molecule, which signals the possibility of treating a series of tumors with immunotherapy, primarily melanoma, lung cancer and bladder cancer. “By doing so we have recovered 20% of patients affected by lung cancer that the human eye had excluded from immunotherapy – adds Dei Tos -. We know that this system is useful for prediction, for example in the response to immunotherapy, it allows to improve the selection of patients and consequently the effectiveness of therapies and thus we will have a prediction in the response to drugs as well as new prognosis parameters”.
The first step is to work on tumors, but algorithms valid for other pathologies could be created. “We are testing 5 protocols for breast and prostate, with a thousand tests already performed in a month and a half and surprising results if you consider that we were expecting a few more problems now the goal is to impact the big killers. There are possible applications in cardiology, hematology and neurology”