AI in medicine, the future is today

Some studies have shown that we will be able to identify more accurately many diseases thanks to the support of artificial intelligence systems, able to diagnose melanomas and predict the risk of heart attack

“Marcus is the new intelligent ultrasound system that thanks to its AI is able to make accurate diagnoses in less than 60 seconds. During the acquisition of the images, Marcus compares the result of the ultrasound with all the hundreds of thousands of medical records present in his cloud, checks the patient’s previous health status, his medical history and indicates the diagnosis and subsequently chooses the cure “.
It is not the beginning of a science fiction story but one of the scenarios that Impactscool brings to schools and universities with the aim of stimulating the debate on the impacts that new technologies will have on our society in the near future. A future that, according to a study published in Annals of Oncology, is already today. A team of German, French and American scientists has developed a software, similar to the human brain, able to identify melanomas with a minimum margin of error, that has proved more accurate than 58 dermatologists from 17 countries. Starting from the images of 100 doubtful cases of skin cancer, in fact, the artificial intelligence system was able to identify 95% of the melanomas, against 86.6% of the doctors.

Before the test, the neural network was subjected to over 100,000 images of skin moles and melanomas, together with the correct diagnosis. In this way, the software has elaborated information and has learned to distinguish, from one photo to another, the dangerous cases from the harmless ones. The repercussions of this research, for now with a symbolic value rather than a clinical one, could be very important: according to data from the World Health Organization, 132,000 new cases of melanoma are diagnosed every year. If the problem is identified too late, when the cancerous activity is advanced, the cure can be decisively complicated. If, to date, there is still no valid alternative to clinical examination, this study highlights how technologies could come soon to the rescue of man: thanks to 2D and 3D photography of the body, in fact, it is possible to capture from 90% to 95% of the skin surface and the evolution of these technologies could lead to an “almost automatic” diagnosis of skin tumors.

But melanoma is not the only pathology that can be identified thanks to the support of an artificial intelligence: even for the infarct, for example, some studies have shown that the machines could provide doctors with fundamental help. To hypothesize the risk of cardiovascular disease, in fact, many specialists based their choice on some guidelines that provide for precise risk factors, including age, cholesterol and pressure levels. Analyzing hundreds of thousands of cases, however, artificial intelligence software is able to create its own “manual”, which can take into account several factors and which would allow to identify a greater number of subjects at risk and thus save lives.
We are not so distant, then, as described in another scenario proposed during the Impactscool’s workshops: the one with HRT, a system that allows you to warn the user twenty minutes before a heart attack occurs, alerting also the family members automatically and, in the most serious cases, the nearest doctor.

Medicine could soon experience a revolution, particularly in the field of diagnostics. For a generation grown up in the myth of the “medical drama” of TV, from Dr. House to Gray’s Anatomy, it will not be easy to get used to the idea.