An App That Senses When You’re Feeling Down just won the $100,000 Data Design Diabetes challenge, because it silently analyzes your phone usage data to figure out if you’re upset. For diabetes patients, that could be a lifesaver.

An App That Senses When You’re Feeling Down
Mel B.

Your smartphone is an incredibly powerful tool–one that we mostly waste by just using to make phone calls and check email. But it’s really an advanced bundle of sensors that is with us nearly 24 hours a day, collecting massive amounts of data. Doctors and health professionals are only now starting to understand the opportunity this data can provide. Take a new app that helps silently identify diabetes patients who might be slipping with their treatments.


Almost one in 10 people in the United States suffer from diabetes (and millions more people are on the brink of getting the disease). Having diabetes can often mean developing mental health issues, including stress and anxiety. All of these problems can lead to lapses in managing the disease., the winner of Sanofi-aventis’s $100,000 Data Design Diabetes challenge to develop a scalable, data-driven product that can benefit anyone in the diabetes ecosystem, keeps track of potentially dangerous mental health issues through a smartphone app–and it does it all without user input., a spinoff company from the MIT Media Lab, didn’t originally aim to help diabetes patients. As part of his thesis, cofounder Anmol Madan collected 320,000 hours of data from research participants’ cell phones. The result: Madan figured out what kinds of cell phone use patterns signal the beginning of issues like the flu or anxiety.

So Madan teamed up with two MIT alumni to launch’s smartphone platform, which runs in the background and analyzes call frequency, location, and text messaging habits to figure out when users are under the weather. The app only looks at statistics, so it doesn’t know if you’re at Burger King and calling your mother. But it does know if you’re staying in one place all the time and neglecting to call anybody.

The diabetes application was an afterthought. “We stumbled upon the use case in diabetes. After looking through clinical literature, we saw a correlation between mood and diabetes,” says cofounder Karan Singh.

When a user displays abnormal behaviors, an alert goes out to their caregiver network. “The idea is for psychosocial support. This helps [users] go a long way to stay on their treatment regiment,” explains Madan. “It’s a ‘check engine’ light.”

The core platform and algorithm for the app is ready to go; now just needs to work on “what that last level of user experience will look like,” says Singh. Next year, the startup will run a study with diabetic patients and clinicians to test whether the caregiver alerts make a difference in disease management.


“We’re excited about having a chance to deploy this,” says Singh.

About the author

Ariel Schwartz is a Senior Editor at Co.Exist. She has contributed to SF Weekly, Popular Science, Inhabitat, Greenbiz, NBC Bay Area, GOOD Magazine and more.