CompanionMX

SUB DATTA: So my name is Sub Datta. I'm the CEO of a very exciting behavioral health technology startup called CompanionMx. We are a spin out out of Professor Sandy Pentland's lab at MIT Media Lab and out of a MIT Enterprise company, called Cogito Corporation, in the Boston area.

And what we do is we have a product called Companion that uses objective data from cell phones, specifically around voice, and cell phone metadata to measure behavioral health symptoms. And clinicians use insights from that data to monitor their patients, to risk stratify, and detect changes in behavioral health symptoms early, and, end of the day, really improve outcomes for their patients.

So corporate history of Companion itself-- so the technology was born at the Human Dynamics Lab, which is led by Professor Sandy Pentland at the MIT Media Lab. In the mid to late 2000s, a gentleman called Josh Feast, who's the CEO of Cogito Corporation, built this company-- at the time, it was called Cogito Health-- to really use social signals to quantify behavior.

And at what time out of that core technology came out two specific use cases. One use case is what Cogito has been building on and is doing extremely well on-- is on using behavioral health signals or empathy AI to really make a huge impact in the customer service market. And the other use case has been around health care and specifically for mood and anxiety disorder patients.

Again, startups, you have its ups and downs, right? And that's when you have to ask yourself a very heavier-- is why are you doing this? And, for me, the reason why I'm doing this is very simple, very, very simple.

What has happened in CHF, what has happened in diabetes, for, like, a type 1, type 2 diabetes patient, the reason why you do not have type 1 and type 2 diabetes patients falling seriously sick all the time is there's something called a continuous glucometer, which continuously measures and gives feedback to the clinician and the patient. So they can take care of themselves.

You don't have anything like that in mental health today. So that's what I want to do. I want to see a world where you have those kind of measures in place. So that way, you can identify such things early.

So it is the 21st century, for crying out loud. There is no reason for anybody to have these kind of episodes. I mean, you can have symptoms. But we should be able to manage it so that, when people don't have these, things don't escalate to this point.

SUBHRANGSHU (SUB) DATTA: What does a near-term, mid-term, and long-term look like for the company? So in the near-terms, the best thing is, unlike several companies classically at our stage, we actually have a product that we have data to show that it works. And not only does it work, people actually love to use it. So that combination is usually a good thing.

And in the near-term, there are very clear opportunities with the movement that is happening in collaborative care, movement that's happening towards value-based care, the movement that's happening towards recognizing and bringing in-- bringing behavioral health back into-- back under the microscope to make a difference on, with the market actively moving in that direction.

In the short-term, we are looking at partner-- at bringing our technology to risk-assuming providers, to peers who are looking at these integrated care networks, to the Department of Defense. So-- and to the different hospital systems that are really looking at-- more closely looking at the collaborative care environment. So I think the timing is absolutely perfect from that perspective.

So in the near-term, that's where we are headed. To figure out how we can really bring in our technologies to make a difference to our patients' and our clinicians' life on a day-to-day basis. Bringing the power-- we are helping our techno-- our clinicians and giving them the power and the tools to monitor their patients and to just try to find-- detect changes in the behavior early.

In the mid- to long-term, we are looking at expanding our capabilities and our reach. So this would be both from a geography and a client customer base standpoint. And it would also be from a straight up capability standpoint.

So right now, our products have been proven to work for mood disorder patients. We want to expand and sharpen our tools for anxiety and stress-related disorders. We want to look at geriatric patients, which is another big population. And we also want to look at the adolescent-age population patients which, as we know today, depression is a huge problem, right? It's-- if you look at the data, suicide has become the number two cause of deaths amongst college-age students and adolescent students today. So that's another place we want to look at.

And one piece that I didn't talk about earlier is we won this big piece of work with the Department of Veteran Affairs and the DOD to-- it's a randomized clinical trial for early suicidality assessment of transitioning veterans. So once we have that work done and we have the evidence based on that trend, that's going to open up a whole set of opportunities for us to really make a difference to-- in the civilian side, to address the scourge of suicides that's happening today.

And on the veteran side, as we know, I mean, between 20 and 23 veterans commit suicide every single day. And if you think about it, even one suicide is too many. So we have this technology which can really enable our clinicians, give them the tools to measure and see changes in behavior early so we can make a difference.

So all in all, I think we are sitting on top of very exciting technology which has been deployed in multiple complex health care systems, one on the government side and the civilian side. The initial acceptance of the product has been amazing. And we look forward to making a difference for our patients and for our clinicians.

We have the right mix. We have great software engineers. We have fantastic neuroscientists. We have a very, very deep expertise within the team in health care. So we're covering all the key aspects, ingredients that we need to be successful.

And I find myself super fortunate. I must have done something right in my last life to have been given this opportunity with a team that I love to be around, with very deep skill sets. And the sheer fun-- and a team with a lot of heart, you know?

So those are three ingredients for me that I think need to be in the right proportion for any company to be successful. And we have that.

SUB DATTA: So we're looking at parameter big picture two broad types of partnerships. One would be with the technology companies and pharmaceutical companies where-- and the second one would be with large employers. So, if I unpack the first category which is technology companies and pharmaceutical companies, really companies that are looking to bring proactive measurement of behavioral health outcomes, so we-- and really trying to incorporate voice-based AI for mental health.

So those partnerships would be something that would be interesting for us. If you are somebody in corporate HR, right, and want to have a conversation with me or with CompanionMx trying to see the fit, what would you be thinking about? So the big things I think is, step one, what wellness programs do you already have in place, and what are you trying to do for your employees? The second question I would also ask is what is the culture in the organization to addressing some of these challenges.

Because, let's face it. There is stigma around it. Not everyone feels comfortable talking about it, so how do you address that and what's that culture within the company to talk about it and to really help out individuals? So, if it's a positive on those two elements, then we can think about figuring out if and how we could bring this proactive measurement and monitoring element that they could work with, and really pull in with clinicians that are already under care to make a difference for their patients.

If you look at the history of CompanionMx, there's a lot of a MIT, right? So, going forward, what can MIT offer us to help us even more? And I think it boils down, I mean, in multiple ways. One is where we're already starting to see some of these benefits. Just going to the MIT management conference last week was extremely helpful.

We got to talk to several partners with whom there might be a fit, so one is giving us through the STEX program through the ILP program getting in touch with the right individuals. And the right organizations having that exposure, and being in an environment where both us and the partner company are interested in talking to each other. So that's sometimes half the battle. That's one.

Two is the being part of the MIT ecosystem, as an MIT alumn and also as a company as a technology that came out of MIT that we are working with, being part of the ecosystem is extremely helpful because this enables us to think about where else we want to go. So, for example, health care is a very big topic at MIT today. Health care analytics is a huge topic at MIT today, and really staying close to that helps us understand as we build from a data science standpoint, from our technology capability standpoint.

As we think about that, it helps us really tap to the right thinking points, right viewpoints, our right technologies early, so that way we can leverage some of those technologies where appropriate. And really shorten the time between an idea and making a difference to a patient's life rate, so that way we are able to do by being part of the ecosystem. And the third part, I would say, is straight out from being part of the ecosystem is we will be building out our team. We'd love to get more MIT alumns in there, and that's going to be across the board.

Once again, I'm obviously a little biased being in MIT alumn, but MIT is hands down the best institute in the world. It has great students, so I would love to tap into that network to build out our team.

SUB DATTA: So the product, Companion, how it works? It's a system. There are three parts to the system. So the first part of the system is what we call a Companion app, which sits on the cell phone. And it takes two broad streams of data. One is voice, which is where the patient or the user would talk to the phone and record something that we call an audio diary. So that's actively collected voice. And the other pieces of information are passively collected cell phone metadata.

So once the app collects those two pieces of information, it goes to the second part of the product, which is what we call the Companion AI, which is in the cloud. So the Companion AI uses our AI-based patented algorithms to convert the two data streams that we talked about, and converts them into four very specific measures of behavioral health symptoms. And those are mood, physical isolation, social isolation, and fatigue. And those are straight from DSM, which is what clinicians are qualitatively trying to evaluate. So we are having quantitative data evaluating and measuring those symptoms, and not only providing scores on the symptoms, but also providing trends of how those scores have been changing over time.

And this piece of information goes to the third part of the system, which is the Companion dashboard, where clinicians can see those pieces of information. And the same information also goes back to that so that the patient can see the same data. So what is happening is at end of the day, you're having a closed loop of information. And as a result of this, there's a higher engagement, there's greater visibility on a continuous, repeatable, and reproducible manner of changes in symptoms of behavioral health of the patient. So that way, the clinicians can detect them early and do something about it.

Both of them. So it looks at call log, text log, and geolocation, so basically a classic cell phone metadata. And the key point to be noted is looking at patterns. So for example, in the voice bit, it's non-word. So it's looking at speech patterns versus the words spoken themselves, which makes it extremely powerful, which makes it very difficult to game, and makes it language agnostic.

And similarly, on the metadata, it's looking for patterns on the frequency, diversity, et cetera, et cetera, basic classic usage patterns, which is what we call also as behavioral biomarkers, and the widespread, we call them acoustic biomarkers. So we're taking these acoustic biomarkers and behavioral biomarkers just like what happens in any other clinical intervention today or any other problems, or like we do when talking about diabetes, right?

So we're taking those measures and then converting them into a vital sign for behavioral health, which is what Companion does, which is converging those four measures that we talked about, physical isolation, social isolation, mood, and fatigue.

Yeah. So how does a clinician use this? So we have used the product in several collaborative care settings. There, a social worker would or a licensed behavioral health counselor would look at the dashboard. And the dashboard, the scores are color-coded, so they're able to see the individual scores across those four measures that I talked about. And they can also see the changes, so they can see trend lines of how things are changing.

So based on a combination of the scores and the changes based on the trend lines, the clinicians make their clinical judgment as to at what point should they engage with the patient. So in several collaborative care settings, they do some collaborative care settings, they do a daily huddle, where a care manager, the PCP, and the behavioral specialist will get together and look at scores. In some places, they do it weekly. So it depends on the current workflow. So the key point there is what we've done in Companion is it's been designed to fit in the current workflow without creating additional work or additional burden for the already overstretched resources that are there in health care.

SUB DATTA: Hi. my name is Sub Datta. I'm the CEO of CompanionMx. We are an exciting behavioral health technology company spun out of the MIT Media Lab and Cogito Corporation, which is a MIT Enterprise Company.

Our solution companion uses a combination of various cell phone data, which include cell phone metadata and voice. And we use our proprietary patented algorithms to convert them into repeatable and reproducible measures of behavioral health symptoms. Used by over 1,500 patients at multiple department of veteran affair clinics and Harvard teaching hospitals, we have shown that clinicians use insights from the measures that we provide through our solution to them. They use them to monitor their patients, to risk stratify and detect changes in behavior symptoms early, and end of the day, cultivate elements that hits the road, really make a difference to patients' outcomes.

So once again, that's who we are. We are CompanionMx. We have a solution that enables clinicians and patients to better take care of themselves, predict onset of episodes, detect changes in behavioral health symptoms, and have a meaningful change, meaningful difference to their behavioral health outcomes.