Abhinav Parate and Akshaya Shanmugam co-founded Lumme Labs to develop products that learn people’s baseline health and habits in order to predict disease and offer recommendations for healthier living.
Parate developed the concept of Lumme while completing his computer science graduate project with Deepak Ganesan, UMass Amherst professor of computer science. Shanmugam joined in after completing her graduate studies in electrical and computer systems engineering. The team is devoted improving the health of Lumme users by analyzing their everyday behavior and making behavioral suggestions -- sometimes, even before poor health develops.
“Lumme is focused on developing an automated software platform that can help us gain greater understanding about our context and how things we do on a daily basis affect our health,” Shanmugam said.
Currently, the Amherst-based lab is working to develop devices and software platforms that can detect addictive behaviors, predict relapses, and offer intervention strategies.
First out of the lab is a system that will help smokers quit by detecting smoking behaviors, and even predicting them before they occur. System development is being funded by the National Cancer Institute, and prototypes are being tested by the Yale University School of Medicine’s Tobacco Treatment Clinic.
“We all make bad health choices,” Parate said. “Smoking, alcohol consumption and overeating are among the primary reasons of preventable death.”
The Lumme platform for smoking cessation consists of a bracelet with embedded sensors that detect certain motions, like raising a hand to take a drag of a cigarette. The bracelet passively detects these motions all the time, and works with software in a smartphone that sorts through the data to learn what a smoking gesture looks like and when it is necessary to intervene.
“Think about everything we do with our hands every day. It’s like finding a needle in a haystack,” Parate said. “Even if someone smokes 20 cigarettes a day, that only translates to about 400 hand gestures.”
Once the user sets their quit date, the smartphone app analyzes data collected by the bracelet about the user’s smoking habits to determine high-risk triggers for relapse, such as the user’s location, time of day, the type of music they listen to, and who they’re with when they smoke.
“Smoking is a very habitual pattern. People smoke with their morning cup of coffee, when they drive back home, or when they are with their smoking buddies,” Shanmugam said. “You can identify patterns and predict when a person is most likely to relapse.”
As Lumme gathers more information about the user’s habits, it employs predictive algorithms to anticipate a potential relapse and send intervention strategies directly to the user’s phone.
These interventions can range from recommending the user avoid their usual ‘smoking buddies’ or take a different route to work. The success of these recommendations relies on what Shanmugam calls “enriched personalized info.”
“If I collected that information and not only said ‘Here’s everything you did today,’ but also gave you personalized tips and strategies you could use to improve your health, that would dramatically change the way we live our lives,” Shanmugam said.
The Lumme system is designed to share information with a behavioral health expert to help give individualized treatment to the user. Currently, the platform uses clinically validated therapy knowledge that Sherry McKee of the Yale School of Medicine has used in her practice for over 15 years.
After several rounds of clinical trials through the Yale University School of Medicine, Lumme can detect smoking behaviors with 95 percent accuracy, and predict a potential relapse up to six minutes before it happens.
Lumme will launch a pre-order on Dec. 18, 2018, Shanmugam said, equipped with highly accurate motion tracking, personalized cognitive-behavioral interventions and a web platform for behavioral experts or health coaches to interact with the user based on the information collected.
The app itself is meant to be fully automated, in concert with supplemental treatments from health professionals. Shanmugam said she would eventually like to scale the product to be useful to individual users without a need for a behavioral expert. She also anticipates extending the product to help manage other addictive behaviors like alcoholism and eating disorders, as well as to monitor chronic diseases, track postoperative recovery progress and diagnose diseases early on.
In May 2017, Lumme Labs won $10,000 in the Valley Venture Mentors Accelerator Awards for area startups. Shanmugam credits the team’s entrepreneurial spirit to their time at UMass, which she and Parate say offered them countless opportunities for collaboration. Without that experience, working intimately with computer scientists, engineers, psychiatrists and medical experts would not have come so naturally, she said.
Shanmugam, who was named this year in Forbes’ 30 Under 30, added that aside from their diverse expertise, the team shares one thing: the spirit of entrepreneurialism.
“Share the passion to change the world,” Shanmugam said. “You have to be a little eccentric to be an entrepreneur.”
Written by Morgan Hughes