Eugenia Chock, MD, MPH, aims to improve the care of women with autoimmune conditions. Current screening for these disorders often results in delayed care, she says.
An assistant professor of medicine (rheumatology, allergy and immunology) at Yale School of Medicine (YSM) and Yale Center for Clinical Investigation Scholar, Chock researches maternal health and offspring outcomes among patients with rheumatic diseases. She is interested in utilizing large clinical datasets to support her work.
Recently, Chock received funding from the Yale Center of Excellence in Regulatory Science and Innovation-Food and Drug Administration Office of Women’s Health to develop the use of AI to remove barriers to diagnosing and addressing autoimmune diseases in women.
In a Q&A, Chock discusses why early diagnosis of autoimmune conditions is important, how machine learning can help, and her hopes for the future of artificial intelligence in medicine.
Why is it important to diagnose autoimmune diseases in women?
In the U.S., approximately 50 million people are affected by autoimmune diseases, and the number is rising. The incidence of systemic lupus erythematosus, for instance, has nearly tripled in the U.S. over the last 40 years.
Eighty percent of individuals affected by autoimmune diseases are women, and many of these diseases have systemic implications, meaning they involve multiple organs. Sex differences influence the onset and severity of these diseases, which can be fatal. Timely diagnosis and treatment can ensure optimal outcomes.
Tell us about your novel machine-learning approach to improve diagnostics in this area.
Many patients are referred to a rheumatologist because they receive a positive antinuclear antibody, or ANA, test result. An ANA test is a very common blood test that screens for autoimmune diseases, particularly lupus and scleroderma. But this test is not perfect. Not all individuals who test positive for ANA have or will develop autoimmune diseases.
Since many people test positive for ANA and are referred to a rheumatologist, the wait lists for these specialists are long. This is a disservice to patients who do have or end up developing an autoimmune disease because their prompt evaluation and care are delayed. In addition, getting a positive test result can cause unnecessary fears among patients, especially if they have to wait a long time to see a specialist.
I am working in collaboration with Na Hong, PhD, instructor of Biomedical Informatics and Data Science at YSM, to use artificial intelligence software to efficiently extract data from electronic health records. Using a machine-learning tool, we’ll identify patients who test positive for ANA and are at risk of developing autoimmune diseases. We’ll also identify patients who have an ANA and don’t develop autoimmunity. All of this is done confidentially in a secure environment. Once we have these two groups of people, we’ll find data points—such as lab test results or medications—in the electronic health records within the Yale ealth system that indicate whether a patient has additional risk factors to develop lupus, scleroderma, or another autoimmune disease down the road.
Once we gather this information from the electronic health records, we’ll apply artificial intelligence software to create an algorithm to help us identify ANA-positive patients who are at higher risk for developing an autoimmune disorder.
What do you hope to accomplish by using AI in medicine?
Currently, we receive many referrals for patients who test positive for ANA, and we don’t know which ones are high risk. I’m hoping that this tool can accurately help physicians identify people—especially women—with autoimmune diseases early on and ensure that they get the appropriate care.
AI in medicine is in its infancy stage. While AI holds much promise, it’s not yet sufficiently refined or reliable to help diagnose or manage medical conditions. It’s just not that sophisticated yet.
My goal is to develop this tool carefully and intelligently to improve the lives of patients and to contribute to the long-term advancement of technology in rheumatology. Once validated by testing in real-world clinical practice, we hope to apply AI for autoimmune disease screening more broadly, improving health care in many health systems nationally.
Source: Yale University