From Developer to Successful Machine Learning Entrepreneur: David Moss, Co-Founder, President and CTO of People Power Company (Part 4)

Sramana Mitra: When did this insight on senior care come to focus?

David Moss: We knew we wanted to get into senior care around 2016. It takes time to orient that product in that direction. We had to keep the company alive. We wanted to position this technology more in this direction.

It was in 2018 that we received a $4.5 million grant from the National Institute on Aging to create assistive solutions for people with Dementia and Alzheimer’s. That was key. That enabled us to focus on commercializing this infrastructure and go out there with something that is valuable for this population.

Sramana Mitra: How long did it take you to build what you needed to build?

David Moss: From a platform and technology standpoint, it was many years. To orient this in the direction of senior care was not a short amount of time. As an AI and IoT company, we are using sensors inside people’s homes to understand their lifestyle patterns and to detect falls, predict health problems, alert caregivers. We are doing so in a very private way.

Even though we had this popular camera app, cameras aren’t that popular in the homes of these seniors. We had to rethink that a little bit. We spent a long time trying to explore the features and the user experiences that would be intuitive for this type of population. What we are targeting is not the seniors themselves; we’re providing solutions for the caregivers.

Sramana Mitra: Are you familiar with Dina Katabi’s work at MIT?

David Moss: I’m not.

Sramana Mitra: MIT does this alumni day. At one of those, Dina spoke about similar solutions that her lab was working on. Going back to your solution, what is the architecture? Originally, you were repurposing an old iPhone to act as the camera.

At this point, I imagine you have a different configuration. In that IoT/AI configuration, what are you doing? Are you doing image processing? Are you detecting falls through image processing? What are the actionable alerts?

David Moss: Let me start at the application and then I can describe the platform and infrastructure that enabled that. We are able to detect falls in real-time with no cameras, buttons, and wearables. We do that in a couple of ways. On my camera here, I have a sensor on my ceiling. It’s a radar-based sensor. It connects to our platform. We have other sensors.

We are able to bring all of these together and form solutions that can get deployed into any type of living space. We’re doing falls in real-time. You wouldn’t traditionally use neural networks for the type of machine learning you’re doing here. There’s a whole class of algorithms and a way of thinking about time series machine learning.

We’re able to understand people’s lifestyle patterns and know when those patterns break unexpectedly. If grandma walks into a hallway and doesn’t come back, there’s a false safety net here provided by these algorithms. These machine learning algorithms are solving a money problem. These sensors cost money. If they didn’t cost money, we can just populate everything with the sensors.

The fact that we can’t means that we’re dealing with sparse data and, sometimes, no data. Machine learning algorithms are there to help fill in the gaps. We can tell, for example, if a person leaves their home and a spouse or loved one is still in bed. Even though we have no data from the bed, the machine learning algorithm can predict that there’s still somebody home even though we can’t see them.

One of the things that get people thinking about senior care in the first place is, a person will have a fall and the family needs to think about dealing with that. The second is understanding people’s lifestyle patterns. This is called activities of daily living. We need to understand people’s mobility, their bathroom habits, their sleep scores, their meals, and their medication.

If these activities trend in a direction that’s abnormal, this may be a cause of concern. This may help uncover hidden health problems. This also delivers situational awareness for people who are remote like a loved one who lives far away from their mom or dad. They can scratch that itch and know everything is okay.

This segment is part 4 in the series : From Developer to Successful Machine Learning Entrepreneur: David Moss, Co-Founder, President and CTO of People Power Company
1 2 3 4 5 6