A Raleigh start-up that has created an AI-based system to collect data and offer advice on the treatment of early-stage Alzheimer’s disease says it’s now working to expand the number of clinics it works with.
CEO Vik Chandra and chief technology officer John Walker, co-founders of uMethod Health, say its system is able to combine the latest research in the field with data about a patient’s genetics, demographics, bloodwork, medical history and lifestyle to help doctors figure out what might be contributing to the patient’s cognitive decline.
The system also can help doctors come up with a treatment plan, they say.
Like cancer, Alzheimer’s is a complex disease with multiple causes that doesn’t lend itself to a single treatment, Chandra said. And there aren’t any medications that offer anything like a cure for it.
“What the researchers are coming to the conclusion on is that the traditional way of trying to address Alzheimer’s just does not work,” said Chandra, a N.C. State University alumnus and former IBM executive with multiple business start-ups to his credit. “The disease is just too complicated to address through a single treatment.
UMethod’s system has been under development for five years and the company has been working with patients for the last three. The system is now in the hands of about 100 care groups across the country.
“This is commercially available at this point in time, and we’re working to scale the commercial availability,” Chandra said, adding that while the system isn’t regulated by the U.S. Food & Drug Administration, it does qualify for Medicare reimbursement.
The idea of applying artificial intelligence, machine-learning systems specifically, to health care isn’t new. Nor is the idea that such a system can help doctors come up with a diagnosis or treatment plan. IBM has gotten a lot of publicity, good and bad, for its attempts to find uses and a market for its Watson AI platform in oncology, for example.
For the moment, Chandra and Walker think uMethod’s system is currently the only one like it in the Alzheimer’s arena. It’s of enough interest to researchers that the company has been able to present findings in each of the past three years at the Alzheimer’s Association International Conference. The most recent just concluded in Chicago.
The company has worked with doctors from the Cleveland Clinic, the Mount Sinai School of Medicine, Duke University and Cornell University as it’s developed the system. And it’s attracted investment from the likes of The Pink Ceiling, an investment group pharmaceutical entrepreneur Cindy Eckert founded in 2016 to back start-ups either launched by or that seem likely to help women.
“I think it surprises most people to learn that two-thirds of Alzheimers’ diagnoses are [of] women,” Eckert said in an email. “Left unaddressed, Alzheimer’s will not only cripple families economically, it will cripple the health care system as we know it. Instead of continuing to wait for a magic bullet to fix this epidemic, we must draw on all of today’s scientific information and create comprehensive personalized treatment plans to slow cognitive decline.”
Walker, a UNC-Chapel Hill alumnus and former IBM researcher, said UMethod’s system is particularly useful in in rooting out drug interactions that can contribute to depression and other problems that make things worse for patients.
The ultimate test of a treatment plan is how well patients do on the common tests doctors use on people over age 65 to see whether their cognition is declining, holding steady or improving, he said.
UMethod also has to make sure it’s “able to explain [to doctors] why our system is recommending something,” Chandra said, underscoring that an AI system is there to help physicians, not supplant them.
Conceptually, the use of an intelligent-adviser system to help develop treatment plans makes sense and should be able to help if it can process all the relevant data and risk factors, said Todd Cohen, a UNC-CH researcher who specializes in neuro-degenerative diseases like Alzheimer’s and ALS.
But the key problem is assembling data in a systematic way, he said.
“To me, the major question is how we’re going to get all that information from people” and do so in a consistent, reliable fashion, Cohen said.
Making sure data-gathering practices are consistent from clinic to clinic is a problem, he said.
“This is a very common problem, how do you establish criteria for consistency with different materials, different handlers, different clinicians and different environments,” Cohen said. “The point is to have one common set of rules you can make predictions on.”