Researchers will use AI to predict who might get certain rare diseases

A team of researchers from the University of Florida Health and Penn Medicine is using a set of artificial intelligence-powered algorithms called PANDA to find rare “zebras” in patients’ medical records and help Patients with some rare diseases are diagnosed and treated more quickly.

In the healthcare world, rare disease Sometimes called ‘zebras’ because they are so unusual and unexpected. Any disease that affects less than 200,000 people nationwide is considered rare disease. Worldwide, there are about 7,000 known rare diseases. In the United States, the total number of people affected by these conditions is about 10%.

According to Jiang Bian, PhD, a professor at the University of Florida College of Medicine, the symptoms of rare diseases are often vague and confusing, and because so few people are affected, diagnosing them can be difficult. towel. University of Florida Medical School scientist.

For this reason, Bian says, “Some patients with rare diseases can go undiagnosed and untreated for many years.” Bian is part of a team of researchers from UF Health and the Perelman School of Medicine at the University of Pennsylvania who are using artificial intelligence and electronic health records to develop an alert system that will alert doctors whose patients are likely to have some rare diseases.

The researchers will develop a set of algorithms supported by machine learninga form artificial intelligence, to determine which patients are at risk for 5 different types of vasculitis and 2 different types of spondylitis, including psoriatic arthritis and ankylosing spondylitis. These predictions, derived from information already in a patient’s electronic health record, can significantly increase the chance of a patient being diagnosed earlier.

Trying to develop this prediction methodcalled “PANDA: Predictive Analysis Through Networked Distributed Algorithms for Multisystem Diseases,” will be led by Bian at UF, and Yong Chen, PhD, professor of biostatistics and Peter A. Merkel, MD, MPH, chair of the department of rheumatology and professor of medicine and epidemiology at Penn.

“This is an exciting step forward, building on our current PDA framework, from clinical evidence generation to AI-informed interventions in clinical decision making,” said Chen. “Despite the clear need to minimize dangerous and costly delays in diagnosis, individual clinicians, especially in primary careface important challenges. “

Chen used one of the forms of vasculitis being studied, granulomatosis with polyangiitis, as an example of the promise the PANDA system holds. This condition involves multiple organ inflammation and can be extremely serious or even fatal. Patient mortality remains high for the first year after diagnosis, and the correct diagnosis of this type of vasculitis, and all others, can be delayed by months or even years.

“Earlier diagnosis of any of the types of vasculitis and spondylitis we are studying will lead to a better prognosis and better clinical outcome,” said Mrs. “Even if we determine that a patient has only a 10% chance of having one of these diseases, the likelihood of a rare problem is much higher and clinicians can keep that in mind and make better decisions for their patients.”

Among the diagnostic challenges faced by clinicians and their patients is how rare diseases can disguise themselves as other common diseases. Clinicians can also be hindered by lack of access to data or other clinicians with which patients work, and simply lack of familiarity with such uncommon conditions. . One algorithm automatically scans known information to determine the likelihood that a disease like GPA could be lifesaving.

“The increasing availability of real-world data, such as electronic health records collected through routine care, presents a golden opportunity to generate factual evidence to support the development of health outcomes,” said Bian. provide information for clinical decision-making,” says Bian. “However, to take advantage of these large real-world collections of data, which are often distributed across multiple sites, new distributed algorithms like PANDA are needed.”

The researchers plan to obtain data through PCORnet, the National Patient-Centered Clinical Research Network. The integrated collaboration of this large clinical research network contains health data from more than 27 million patients nationwide. Identified data from these patients, including laboratory test results, comorbid conditions, previous treatments, and other commonly available information, will be used. to create algorithms. Once built, the researchers will test the predictive power of each algorithm on more than 10 health systems. The methods the team develops will be shared and available to be applied to other diseases.

As their name suggests, machine learning algorithms are designed to “learn” and refine themselves as they are used and provide more data. For this reason, it is possible that PANDA will become more useful as time goes on.

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