Researchers will use artificial intelligence to predict who is likely to develop certain rare diseases

A team of researchers at Florida State University of Health and Medicine is using a set of artificial intelligence algorithms called PANDA to find rare “zebras” in patients’ medical records and help patients with certain rare diseases get diagnosed and treated more quickly.

In healthcare circles, rare diseases they are sometimes called “zebras” because they are very unusual and unexpected. Any disease that affects fewer than 200,000 people nationwide counts a rare disease. About 7,000 rare diseases are known worldwide. In the United States, the total number of people suffering from these diseases is about 10%.

According to Jiang Bian, Ph.D., a professor at the University of Florida College of Medicine, because the symptoms of rare diseases are often vague and confusing, and because there are so few people, diagnosing them can be difficult. scientist at the University of Florida Public Health.

For this reason, Bian said, “Some patients with rare diseases may go undiagnosed and untreated for years.” Bian is part of a team of researchers from UF Health and the Perelman School of Medicine at the University of Pennsylvania that is using artificial intelligence and electronic medical records develop an alert system that will alert doctors whose patients appear to be suffering from certain rare diseases.

The researchers will develop a set of algorithms based on machine learningform artificial intelligence, to determine which patients are at risk for five different types of vasculitis and two different types of spondyloarthritis, including psoriatic arthritis and ankylosing spondylitis. These predictions, derived from information already available in patients’ electronic health records, can significantly increase the likelihood that patients will be diagnosed earlier.

Efforts to develop this forecasting methodtitled “PANDA: Predictive Analytics Using Network-Based Distributed Algorithms for Multisystem Diseases,” will be led by UU’s Bian Yong Chen, PhD, professor of biostatistics, and Peter A. Merkel, MD, MPH, chief of rheumatology and professor of medicine and epidemiology in Pennsylvania.

“This is an exciting step forward that builds on our current PDA structure, from generating clinical evidence to AI-based intervention in clinical decision-making,” said Chen. “Despite the obvious need to reduce dangerous and costly delays in diagnosis, individual clinicians, particularly in primary careface important challenges.”

Chen used one of the forms of vasculitis under investigation, granulomatosis with polyangiitis, as an example of the promise of the PANDA system. This condition involves inflammation of many organs and can be extremely severe or even fatal. Patient mortality rates remain high in the first year after diagnosis, and the correct diagnosis of this type of vasculitis, like all other types, can be delayed for months or even years.

“Earlier diagnosis of any of the types of vasculitis and spondyloarthritis that we are working on leads to a much better prognosis and better clinical outcomes,” Merkel said. “Even if we determine that a patient has only a 10% chance of developing one of these conditions, that’s a much higher chance of a rare problem, and clinicians can keep that in mind and make better decisions for their patients.”

Among the diagnostic challenges facing clinicians and their patients is how rare diseases can masquerade as other common diseases. Clinicians may also be deterred by lack of access to data or other clinicians with whom the patient is working, and simply not familiar enough with such unusual diseases. An algorithm which automatically scans known information to determine if a disease like GPA can be life-saving.

“The increased availability of real-world data, such as electronic health records collected during routine care, provides an excellent opportunity for real-world evidence to inform clinical decision-making,” Byan said. “However, to exploit these large collections of real-world data, which are often distributed across multiple sites, new distributed algorithms like PANDA are sorely needed.”

The researchers plan to access the data through PCORnet, the National Patient-Centered Clinical Research Network. This integrated partnership of major clinical research networks contains health data on more than 27 million patients nationwide. De-identified data from these patients, including laboratory test results, co-morbidities, past treatment and other publicly available information, will be used to create the algorithms. Once established, the researchers will test the predictive ability of each algorithm in more than 10 healthcare systems. The methods developed by the team will be general and available for application to other diseases.

As their name suggests, machine learning algorithms are designed to “learn” and improve as more data is used and transmitted. For this reason, it is possible that PANDA will become more useful over time.

“Ultimately, we hope to build on algorithms developed for rare diseases and apply them to other diseases,” Bian said.