Healthcare resource platform uses AI to filter results
GetWellAI informs individuals of alternative treatment options
Using Google to self-educate about a chronic illness is not the easiest platform to navigate nor the most-informed choice for patients, argues Subhankar Ray, founder of the healthcare platform GetWellAI, because what appears on the first page of results is not always the best treatment option.
On GetWellAI, whether it’s ADHD, Crohn’s Disease or lower back pain, all treatment options are laid out in ranking order, leading to portals of information about each of them.
“Our software looks at different websites, journals and clinical studies and does some analysis to understand how the treatment plans are, compared to other treatment plans,” says Ray. “If we see something in scholarly journals and clinical trials, it is weighted higher. We have information about clinical trials that are not happening here but are happening all over the world.”
Intelligent data gathering
GetWellAI gathers some information through partnerships with medical associations, but mainly the data is gathered from its intelligent, mission-based “crawler,” using text analytics.
“What is gaining momentum [in the high-tech industry] is a mission-based learning artificial intelligence. That’s what we use,” says Ray. “It’s using machine learning techniques to learn from its mistakes so it can crawl the internet efficiently.”
How exactly does AI learn?
As Ray explained, in very simple language, an AI-based machine learning system creates a model or equations based off of past data. A model could be as simple as a linear function (a straight line), which is never the case, or a complex, non-linear function.
Creating a model requires finding out the value of the constants — those constants, together with the independent (x) and dependent (y) variables, create an equation or model. To use a model to predict a dependent variable (y), a new value (not included in the training set) is plugged in to get a prediction of the value for the dependent variable ‘y’.
“We’re getting better, and lucky, with new algorithms to create models coupled with new approximations in the generation of models using convoluted neural networks,” says Ray. “Google and others are creating algorithm-specific dedicated hardware to do complex calculations faster. This brings better vision with quicker perception to a machine and that’s the reason we’ll be seeing self-driving cars, general purpose robots learning quickly from watching a video or from each other, or humans doing any repetitive task.”
Providing a benefit
With an artificial intelligence scanning and evaluating healthcare materials for users, Ray argues the platform holds value to employers seeking to make their workforce healthier and lower healthcare costs. While he encourages users to discuss alternative treatment options with their physician, Ray sees promise in connecting workers with chronic illnesses to less expensive and more effective treatments.
“Everybody’s genetically different so some treatment plans that work for some, may not work for others,” says Ray, who is currently working with a professor at MIT focused on bioinformatics. (GetWellAI is jointly-based out of Cambridge and the Manchester, NH-based incubator Alpha Loft.)
GetWellAI could also be a product offered by wellness companies, as their "missions align well," says Brian Murphy, who is part of the GetWellAI team and previously worked with healthcare brokers. He points out that most wellness programs are focused on preventative care — GetWellAI would provide remedies for those who have a chronic illness.
To ensure employers aren’t paying for a product that’s not utilized, the platform will charge a monthly fee per user only if the employee logs in, with no additional charge for how often the platform is utilized.
Ray maintains the platform adheres to HIPPA privacy requirements and does not share protective health information with the employer.
“It’s informing individuals so they can make their own informed decision,” says Ray of GetWellAI. “Our goal is in the medical world to find things that are good but not popular yet.”
The company is currently seeking beta testers for the platform. Anyone interested is encouraged to contact Subhankar Ray at firstname.lastname@example.org.