Q: Let’s start with autoimmune diseases themselves. It seems like there are many of them—and they affect many people.
LIU: There are around 80 relatively well-known autoimmune diseases. Many of them are quite rare, but collectively, autoimmune diseases are fairly common. Around 10% to 20% of the population are affected by some kind of autoimmune disease.
Q: Many autoimmune diseases seem to share the same symptoms.
LIU: That’s right. Clinicians can often tell diseases apart, but often the diagnosis takes a long time because there are a lot of shared, nonspecific symptoms between various diseases. And because of that, people who otherwise are generally healthy might not turn to a rheumatologist, even though they might be at greater risk for autoimmune disease.
Q: Is it dangerous to ignore those symptoms?
LIU: Those symptoms can, unfortunately, cause irreversible, behind-the-scenes damage to the body. But the most clinically actionable cases are the ones showing early, nonspecific symptoms. That’s why it’s very important to be able to figure out what disease you’re at the risk of developing to get specialized, tailored treatment. Early diagnosis and intervention lead to improved treatment and better disease management.
Q: You’re using AI to build a tool that would better track disease progression. What does that look like?
LIU: We used AI to study the electronic health records—originally developed for billing purposes and insurance company reimbursements—to look at the health history of a large number of individuals to identify those with preclinical symptoms of autoimmune disease. We then analyzed how many of them actually developed an autoimmune disease, and we also use AI to analyze large genetic studies of people with autoimmune diseases to come up with a model that can predict disease risk. AI and machine learning play a critical role here, because the datasets are huge and finding different combinations of genetic mutations that can actually predict the disease risk is very challenging.
Q: Once you’ve fully developed this tool, will rheumatologists be able to use it?
LIU: Yes, but eventually and under proper guidance, I think patients themselves would be able to use it. If they have autoimmune disease symptoms, all they’d need is to send in a saliva sample with a DNA genetic testing kit, use our tool to calculate a risk score to see how likely they are to develop a disease, and then seek medical advice immediately.
Dajiang Liu is a distinguished professor, vice chair for research, and director of artificial intelligence and biomedical informatics in the Penn State College of Medicine.