The technology is already being used in medical centres across Europe and beyond, according to the scientists behind it.
Developed by the medical-tech startup Saventic Health, the system uses AI algorithms to flag patients whose symptoms may indicate a rare disease, significantly shortening the time to diagnosis.
While the system does not provide a final diagnosis – that remains the job of the physician – it can alert doctors to the possibility of conditions that are often overlooked.
"Our algorithms don’t diagnose diseases. That’s still the doctor’s role," said Dr. Marek Dudziński, a medical doctor with Saventic Health. “What we do is identify patients with a high risk of a particular disease and point them toward the right diagnostic path.”
In the European Union, a disease is classified as rare if it affects fewer than five people per 10,000. Ultra-rare diseases affect just one person in 50,000 or fewer.
Some conditions are so rare that only a handful of people worldwide are affected.
Despite their low individual prevalence, rare diseases as a class are surprisingly common, affecting around 450 million people globally, including an estimated 2 million in Poland.
The challenge is that there are around 8,000 known rare diseases, and symptoms are often vague or resemble more common conditions.
"You can’t expect doctors to be aware of all 8,000 rare diseases," Dr. Dudziński said. As a result, patients typically wait five to eight years for an accurate diagnosis.
Saventic Health’s algorithms can help change that.
In one high-profile case, published in the medical journal Blood, researchers from Poland and Brazil used the system to analyse nearly 600,000 medical records from a hospital in Brazil.
The AI identified 102 patients with possible signs of Castleman disease, a rare condition affecting the lymphatic system. Of those, three were flagged by Brazilian doctors for further evaluation, which is still ongoing.
Dr. Karol Lis of Saventic Health said the company’s technology is uniquely valuable because it analyses a patient’s entire medical record.
"Almost no one else in the world does this. Many companies work with image data, like CT or MRI scans, or just parts of the medical record. That’s much easier," he said.
According to the team, the system integrates available AI techniques and focuses specifically on diseases that already have some form of treatment available.
It is already in use in Poland, France, Germany, Sweden, Austria and Switzerland, and has also been deployed in Canada and Brazil. The company is now expanding to Taiwan.
(rt/gs)
Source: PAP