TWIDH - Emilia Molimpakis, PhD - CEO & Co-Founder at thymia
We had a chance to chat with Dr. Molimpakis about her startup, thymia.
What is thymia?
The current system for assessing mental health issues, and in particular depression, is based on questionnaires that are subjective and prone to bias. In addition, there is no standardized way to monitor patients at home. What's more, clinicians frequently complain about spending too much time on admin tasks and not enough time with their patients.
thymia uses video games based on neuropsychology and analyses of facial microexpressions & speech patterns to make mental health assessments faster, more accurate, and objective.
thymia's end-to-end solution works from the moment the patient first visits the clinician through to the end of treatment, empowering clinicians with more information, saving them time and costs while improving workflow and patient outcomes.
(edited for publication)
Good afternoon and welcome to This Week in Digital Health. I am your host, Mike Doherty. And we are with Dr. Emilia Molimpakis. She is the recipient of the Young Innovator Award for 2021. thymia was named one of the top 21 companies in mental health worldwide. And by background, she has a Ph.D. in Neuroscience and Linguistics, which I probably mispronounced as well. I know that you were a video gamer and scientific researcher. So that's a lot going into that. So tell us a bit of what led you in this direction?
It was such a smooth pathway; I'd say it's not what I set out to do, essentially. But we found a female, me and my co-founder, about a year and a half ago now. But before that, I was a researcher for quite a while; I worked in academia for roughly 12 years. So I did a Ph.D. and postdoc at UCL, and my specialty was cognitive neuroscience and linguistics. And essentially, I worked for all that time with various patient populations. Specifically, people with mental health issues and cognitive disorders started with Alzheimer's disease and Parkinson's and then moved through to mood disorders, including schizophrenia and depression. But for all of them, I would always approach them in a very similar way, which is my research focused on how we can use language as a biomarker for their cognition. And by language, I mean both the acoustic properties of how somebody sounds, so you know, the physical sound waves and the content of what they say. So the semantics, syntax, etc. And these are interesting and good handles for you to understand how somebody is thinking and whether everything is working as we would expect it to be, or if there are any issues, essentially, when I was doing my postdoc.
Towards the end of my Ph.D., I worked with the video game company on the side, which is quite a small one. But I was doing exciting projects, like helping them develop this video game where the game levels got progressively harder as you went on. And it was based on neuroscience protocols - very interesting as a concept.
However, while I was doing my postdoc, and working with this company, my best friend, also doing a Ph.D., developed depression. This is quite common with academics. And like a lot of people, she tried to go through the UK mental health care system. But unfortunately, again, like many people, she fell through the cracks in the system. And so though someone was seeing her, she had seen her GP, she'd been referred to a psychologist and a psychiatrist. Within just three months, she, unfortunately, ended up trying to kill herself, which was obviously like, as a best friend, very hard to go through because I didn't see how bad it was. But what shocked me was, okay, I'm, you know, not a specialist. And I couldn't see what was going on. But her psychiatrist, or psychologist, surely they could tell that this was this bad. Like it didn't have to get to this point.
And that's what prompted me to start looking into how psychiatrists, psychologists, and even GPS go about diagnosing and assessing and monitoring mental health issues, particularly depression. And I realized that the system is remarkably flawed. There are incredibly subjective assessment systems in place. There are these questionnaires; essentially, there are a few quite well-known ones like the PHQ-9 Depression Test Questionnaire, where you're essentially asking the person, Okay, you say that you feel sad on a scale of zero to four? How sad Have you felt in the past few weeks? Or how tired have you felt like you're leading the person? Very subjective.
And all these studies have shown that when you ask somebody right here and now how they felt about the past two weeks, if they feel bad now, they will retrospectively consider the past few weeks to be bad as well, or vice versa. And so you're not getting the whole picture. So I realized actually, a lot of the stuff I was doing during my Ph.D. in my postdoc, these biomarkers essentially, weren't being used by clinicians. But, still, I could help them by bringing them into the industry.
And also, you know, if I also realized if we tweaked the video games similar to the ones that I was developing, we tweaked them a little bit, we could use them to target patterns that are typical for depression. And so that's where the idea for thymia essentially came. So through personal experience and life factors, I ended up leaving academia. And I joined Entrepreneur First. It's an accelerator program. There are other outposts in; I think there's in Paris and Berlin and various other places. And I was looking for somebody to help you build thymia. And that's where I met Stefano, my co-founder, and he was just the perfect match in terms of skill sets and backgrounds. So that's how thymia was born. Quite long-winded, but it's the whole story, essentially.
And your co-founder, Stefano, has a Ph.D. in theoretical physics?
So it sounds a bit odd, right? So why would theoretical physics be relevant at this point? But the relevance comes from the fact that after his Ph.D., which he did about eight years ago, nine years ago, now, he worked for roughly eight years in large investment banks as a quantitative analyst. And there, he specialized in building machine learning models that combine modalities of data. So different modalities, like text, like sound like speech, all these things. And he was building models to help traders trade on the trading floor, essentially. But the same models that he built there can be used with our data to help assess and monitor depression. So it's a very amazingly overlapping skill set there for us, for me to use to help me build the media, and something exciting.
But he has a specialty in something called ethical or explainable, artificial intelligence; I'm not sure if you're familiar at all with the term, maybe your listeners aren't, so that I can go into it a little bit. So your typical AI, and this is something also that sometimes gives AI and machine learning a bad rap, is that you don't know exactly what's happening in the model, it works as a black box, you put stuff in, and it computes, and then it spits out that and you have to trust the process. What explainable or ethical AI tries to do is essentially cut a window into that black box, so you can understand why the model might suggest something. And so then you can also question it and say, is this actually what we expect it to find. So it helps you deal with many biases that are inherently, unfortunately, always incorporated into these models, like biases against racist biases. Then, again, it's gender and age, and explainable AI helps you detect these and correct them, essentially, which is super important for mental health.
And how was the transition from research and theoretical to, you know, you're out in the real world and building real products? How was that transition for you?
I think it was; it was incredibly interesting, exciting, and also very challenging. It was a very steep learning curve. When I first had the idea of Athena, I didn't understand entirely at that point, like how big it would get and how impactful Of course, I wanted to be super impactful. Still, the scale at which we've grown and the speed and enormous amounts of data we collected have been amazing. And I think that was the amazing thing about transitioning from academia to the industry into creating your startup; the best thing was, from the moment you have this idea. You think of something, and you know, you can make it work to have an implemented and working and gathering data. And you know, then of course, also selling it. The timeframe is so small; it's its amazingly quick turnaround, which was the best thing. The other thing was, of course, dealing with investors and stuff like that, which was not something I knew, and that was not great. But, you know, we navigated that, and it's been, I would say, a year and a half, I would not change any of it. Absolutely. Fantastic.
You know, it's interesting, having been an investor, we're always leery of the academic types, right? Because, in our mind, we're like, you know, that's nice, but, you know, have you ever built anything or sold anything? So how did you overcome that?
So there was a little bias, I think, against the fact that you know, I came from academia. Stefano was from the industry, but he had never built products to sell directly to consumers. He was always within this big, corporate environment. So there was a lot. Have you no apprehension, I guess in the beginning about us, and apprehension, I think this will sound bad. But actually, I did get quite a few investors who were; they didn't like the idea that I was a young woman and the CEO at the same time. So combine that with academics.
Yes, especially women in leadership.
Yeah, especially in tech as well, like, it's very unusual. And I just get that a lot. Like, you know, oh, hey, your co-founders, the guy, why isn't he the CEO, and you're the CTO. Like, that's not how it works. But I think essentially what we were able to show was, look, we can talk the talk, but we can also walk the walk. And this is an amazing project. Look at the traction we have, like, people love this. These are our numbers; these are statistics. These are the data. And we're closing really big deals. So I think, essentially, the attraction and showing them how good our models were, was what changed their mindset. You're not just great academics; we're also great at designing and delivering something.
We were fortunate to emphasize the design from a female's point of view. If you visit our website and if you have the opportunity to interact with our product, it's incredibly beautifully designed. And that was thanks to us, investing a lot of our initial funding into, you know, product design, working with this amazing illustrator and design team, South Africa. And they helped us create this feeling for our product that was calming and engaging, and through iterative feedback from patients and clinicians, we made it even more. So that's everything, the colors, the style, the feel, the movements, everything is made with patients in mind and their feedback. And that's what makes it work. Because they come back and they love it. And that's how you, that's the only way the product can work essentially. So we were fortunate to do that straight away. Early on, I guess I want to say it was insightful and everything. But at the end of the day, it has become like a really big part of feeling now.
This is a diagnostic tool?
I prefer to avoid the term diagnostic because that's not its intention; it intends to provide clinicians with as much additional objective information as possible. Combining the modalities of voice, video, and behavior through video games, we do not want to diagnose and replace the clinician in any way. We firmly believe the clinician has their expertise, and they do what they do very well. We just want to empower them with more data because they need it. And also, clinicians are dealing more and more, particularly psychiatrists, with burnout, and not being able to meet the demand and mental health is, you know, we're going through a crisis period, particularly after COVID. And we want to help them meet the demand; it's not replaced them. So we are essentially more of an assessment and monitoring tool. In the future, we will also hopefully be able to assess quickly whether treatment is working or not sooner than the patient can realize. In time, we will hopefully predict what type of medication different people will respond better to, but always with the clinician together.
And you know that the company, I'm assuming, is intentionally named thymia AI. So the AI aspect appears to garner those insights earlier from the data patterns.
Exactly, yes. So essentially, to explain how thymia works and what we do. When you think of depression and any other cognitive disorder or mental health issue, you're essentially looking at a difference in how this person's brain works. It doesn't mean it's worse or better or anything; it's a difference. And this difference is accompanied naturally by cognitive and behavioral patterns and changes in the normal patterns you would see with people without this. And so what we do is we use the three modalities, assess people's speech, assess their facial expressions, how their eyes move, and their reaction times and behavior in the video games. And through these, we look for these signature patterns in cognition and behavior. And each disorder has its pattern. So this means we're starting with depression, we're looking for the signature pattern of major depression, but we also can expand to any other cognitive disorder. So we've already started expanding to Alzheimer's, Parkinson's, Lewy body dementia, and ADHD, and looking again for these patterns in each one.
And obviously, there's, you know, a scale of those two; you're also sort of looking at where they are in the progression.
Yes, exactly. So we cannot only assess whether somebody does or does not have this particular disorder, let's say depression, but we can also quantify how severe or not the disorder is at this time point and also how it progresses over time. So because our tool works longitudinally, each person becomes their baseline, and we can continuously monitor them. So yeah, we can; the other big thing that we do is not just like producing a category of output. We can also quantify the core symptoms of depression, which nobody else is currently doing because you need all these modalities to do this correctly. So we can quantify fatigue, working memory impairment, changes and psychomotor, speed, attention, and mood shifts through these different modalities.
And does the gaming process become therapeutic itself?
That's a great question. I know there are companies out there doing, like using gaming as a therapy in and of itself, like, there's a Kelly, health who does this for ADHD, for instance, we haven't intended to become part of the treatment process. However, the feedback we've had and what we've seen is when patients use this tool. They use it consistently, and then they can see the output over time; they can see a measurable objective difference in their behaviors in response; let's say treatment affects treatment adherence. So you're more likely to stick to your treatment, which in you know, by extent also means that you'll get better.
And also, you're able to provide insights over some time versus a point in time. So, for example, practitioners typically see them in their office; that's a point in time, like you said, depending on how they're feeling that day. And you're either able to look back over weeks or longer to provide greater insights into how they've been doing?
Exactly, yeah, that's, that's one of the biggest value adds a theme; it's not just that we are providing objective information at a point in time to clinicians. Typically, psychiatrists see their patients once every four to eight weeks. And in between, there's nothing. And if your patient is doing poorly, there's no way you would know unless they reach out to you.
Clinical psychology is slightly different, you'll see them more frequently, but psychiatrists, typically, it's once a month or once every other month. So you're dealing with a case of major depression, you're dealing potentially with something that triggers an episode, let's say, it could be anything, it could be, you know, the person gets fired, or the person is going through a divorce or even less. So hence, like the person, you know, their friend moves away, or something like that, any of those things can trigger an episode. And as a clinician, you would never know. And the thing thymia does is we ask the patient to engage with us doesn't need to be every day is once every few days. And it doesn't need to be all of the different games or all the different activities; this one here, one there, it allows us to keep an eye on them and inform the clinician so that they always have an overview. And when they see the patient after four or eight weeks or even a week, they can look back and say, Oh, I see on this day, you weren't doing so well. You were very tired. Can you talk to me a bit more, a bit more about this? So you have so much more insight?
And with that baseline, you know, will you create alerts when someone is going through a particular episode that you're identifying?
That is the goal. But, naturally, you know, you have to go through all sorts of regulatory processes and become higher levels of the medical device if you want to take on the entire liability if that is happening. But we certainly want to alert the clinician anytime we pick up anything; we also offer through our platform the capacity for the patient to message the clinician or us at any given time. And we also have an option for them to call the NHS helpline or whatever corresponding thing you have in states or Greece or anywhere else, as well.
And there's this platform provided, you know, it's sort of like the gaming community where there's an environment to interact with other people on the platform, besides the practitioners.
That's an excellent idea. We haven't thought of that yet. At the moment, each patient occupies the platform for themselves in isolation, just together with their clinician. The clinician themselves can have an overview of all their patients at any given time, contact them. Still, the patient doesn't interact. And that is an interesting suggestion, and we could think about how to do that in different ways, potentially through the games. That could be very interesting because typically one of them, you know, One of the benefits, is just the idea that I'm not the only one going through this. Yeah, absolutely. That's what the community provides that, you know, that, you know, this is, you know, normal, whatever normal is this is, you know, comments, things like that so that, you know, they can interact potentially with others. So you have raised funding, is that correct? Yes.
So, at the end of April, earlier this year, 2021, we raised 780,000 pounds of funding was $1.1 million. And actually, now we are very serial racing. So yeah, we're raising another round, like sooner than we expected. Because essentially, we have a lot of runway left for the minute we already raised. But we've had so much demand for the product; it's, you know, it's amazing. But you don't want to say that this is you expect this or anything like that. So we have contracts now for over 80,000 active monthly users for 2022. And we're negotiating for the contracts to raise that number to potentially even half a million users. So obviously, we need a lot more people on the technical team and support team to make sure that everything goes smoothly and everything. So yeah, we're rounding up raising again.
And are you going direct consumer? Or do you have partners where you're selling into hospitals or practice practices? Or how does that work?Â
thymia is a b2b business model. Essentially, we go straight to mental health clinics, telehealth companies, mental wellness companies. Each of them has its own set of patients and sells direct to them, potentially in the future. So we may go b2c, but for the moment, it makes a lot more sense to go b2b.
And one of the typical digital health models is, you know, are you offering value-based care? You know, can you save money? Can you change outcomes? And how do you respond to that?
Absolutely. So one of the reasons we're successful as a b2b solution is because we address a lot of the commercial drivers for these clinics. So, of course, every clinic wants to save time; they want to save money. And they want to maximize revenue potential. So thymia, essentially through our assessments before and after the appointment and through the teleconferencing solution we offer, we saved clinicians roughly a minimum of 30 minutes per session.
So if you have like 45-minute sessions or 60-minute sessions, you're increasing the capacity for each clinician to see more patients, they don't have to see more patients, but the capacity is there. And particularly for the clinic, this is good, because as I mentioned earlier, many clinicians suffer from burnout, particularly psychiatrists. So this gives them more breathing space and more capacity to handle more of the inflow. So there's certainly no; there is enough demand for mental health practitioners; it's whether they can meet it. And the other thing we do well, which is a commercial driver, not everyone seems to grasp straight away or understand if you're not within this system.
We can help clinicians stabilize a patient much sooner than you usually would without an objective assessment system. And you may, let's say we're being incredibly cynical. You may ask, why would a clinician want a patient to be stabilized sooner? Indeed, we want them to come back again and again so that we can get more money and more revenue; let's be completely cynical here. But actually, what you want is a patient to be stabilized so that that patient essentially doesn't need to go and find another clinician who will find the proper treatment for them.
Ultimately, the clinician and the patient want the same thing. They want stabilization; they want the appropriate treatment. Typically it takes three years for a patient to find the proper treatment for depression. But it could take up to a decade. We recently had somebody tell us that they were undergoing a potential lawsuit in the clinic because they had misdiagnosed somebody as having depression when they had Lewy body dementia. And this has been ongoing for ten years. So we create value there as well. So even if we want to be completely cynical, which is not our intention, we want to make sure that we are helping the patient and we are helping everyone.
Our goal is to make mental health as openly spoken about and as objectively measurable as physical health. As a result, we can help the clinician save money, which means that they'll use our system.
And does the AI provide indicators for earlier insights, then, you know, we might be in the other environment, or can you see things earlier?
That is one of our goals. So, your average antidepressant medication typically takes, unfortunately, six to eight weeks to be perceived to have an effect by the patient and to be observed by the condition. A little-known fact about some antidepressants is within those six to eight weeks. One of the common side effects is they can make you more suicidal, which is not a desired side effect or outcome, in the best of cases, particularly depression. Yeah. So essentially, what we are aiming to prove out over the next year or so is that we will be able to detect a difference, or a response in treatment within the first two weeks, maybe even the first week, which means if there's no response, you change straightaway.
That's an early indicator so that you can pursue another course of action?
Exactly, yes. And the other like the more ambitious target is to see if Actually, we can use the biomarkers we have early on straight from the first assessment to detect or predict which treatments this particular individual is likely to best respond to, which isn't even further, you know, you're bringing the treatment response even further forward. So you don't have to go through one or two weeks; you go straight away to the right one.
I'm also guessing that what you're looking for is to avoid any acute incidences. And you're looking for indicators that might suggest that they're going in that direction?
Absolutely, yeah. So it's essential to understand that there are good days and bad days with any mental health condition, like physical conditions. And one of our targets is to be able to detect the difference between the two. And to be able to prompt the individual or the clinician to do something about it. So absolutely, yeah. And that's not just true for major depression. That's true for bipolar disorder, is true for autism, for ADHD, there's always ebbs and flows with all of these conditions, and having eyes on it at any given point is so important.
I think you mentioned the idea of stabilizing. There is ongoing treatment, we're not suggesting that that would not continue, but we're looking to avoid the bigger acute events where they end up in the emergency room or something.
Absolutely. Yeah. The other thing that we're also hoping to help with is to detect early depression in people where you're not necessarily expecting it. So people, let's say, who may have gone to the doctor for a different reason, or in mothers after birth, let's say if you use a system like ours, potentially we can detect that. So things are changing, which will also be very helpful for clinicians.
I mean, depression is a common side effect for most chronic illnesses, right?
Yes, absolutely. Yep. That's also one of the reasons that we constantly ask anyone who interacts with our platform, whether we're talking about the patients who were gathering data from or the healthy control group, we always ask to understand what other physical conditions they may or may not have, whether also they have things like asthma, or any of those things, something that could affect the outcome or the output of our models, and things that could affect your voice. So all these things quite often frequently interact, and you have depressive episodes in these individuals. So it's good to understand what you're dealing with, essentially.
And what are the challenges in moving from different conditions from you know, like you're talking about depression is one set, but you know, Alzheimer's is different. So how do you? How do you manage that?
Essentially, this comes from using all these different complex modalities; it means you get a very comprehensive picture and understanding of the individual as an individual. And by extension, you also get an understanding of each of the disorders. So as I mentioned earlier like you have this concept of a signature pattern. So each disorder has its signature pattern. And for us to be able to detect depression, it's imperative that we understand depression signature, but also that we understand the other ones so we can distinguish one from the other. So when you're looking at Alzheimer's, Parkinson's, Lewy body dementia, all of these are have very strong speech indications, very strong indications in terms of body movements, eye movements, facial movements. And these are exactly what our models detect. And we also are using a concept called reinforcement learning, which is quite a new AI concept, which means you can essentially use the models you've already trained with a specific disorder. And you can use those models to bootstrap the next ones, which are targeting a different disorder, which means you can, every time you move to a new disorder, you effectively decrease the amount of data, you need to get the same level of accuracy in your models, which allows us to kind of piggyback on our previous success and move on. And so it's essentially a combination of pre-existing models, understanding the disorder well and understanding the signals that we need to identify, and gathering a very large amount of data to make sure it's properly trained.
And are your services currently available in the UK? Where are you?
Our original market is the UK. We are selling here to private mental health clinics, mental wellness companies, etc. We intended to stay in the UK for a while, but actually, we had a lot of inbound interest from the rest of Europe. So we ended up expanding quite quickly to Southern Europe. We have clinics now in we're using us in Greece, Spain, but also other areas. So Brazil is a very big customer.
There are unique, multicultural issues.
It's not an issue for us because our models are language and culture agnostic. And each person becomes their baseline. So it's more of an issue, let's say if you think about it with accents within the same language, that can also become a big thing linguistically, but because we ensure we gather data from literally 10s of 1000s of individuals to train our models, but we also longitudinally, monitor the same people. So we can control for these factors, essentially.
And when you raise the new round of money, I'm assuming this is expansion and growth; where will you target? New users new geographic?
The aim is, by the time we get to our Series A round, we want to have also expanded to ADHD, bipolar disorder, Alzheimer's, and Parkinson's, and also have expanded to, as I said, Europe, Brazil, potentially, the Middle East, we have a lot of interest there. But we're still evaluating whether or not this is a good idea. So then, we would expand to the States; essentially, we're aiming to raise around our Series A round there.
And what is the competitive landscape look like?
Yeah, so this is also a great question. So we are the only company as yet to combine multiple modalities. Plenty of other people out there have tried to address depression or Alzheimer's, various cognitive disorders with a single modality. Most Typically, this would be a voice. So quite a big competitor in the States, for instance, is a company called ellipsis. Health, they do depression assessments by voice, Sunder health also in the states do something similar. However, when you only have one modality like voice, there are certain issues, essentially, with running your platform in that way. While voice is excellent, and it has a strong signal.
In the lab, in the real world, you can't rely just on voice to make these assessments, particularly not longitudinally. Because quite often you deal with the concept and the issue of noise, both metaphorically And you know, if the microphone isn't working, or a child is screaming in the background, there's a dog barking, it's going to affect your model. But quite often, with depressed individuals, you have the issue that they don't want to talk on a given day, they don't want to open up their phone, and you know, start talking to it. Whereas if you have multiple modalities, your model is a lot more accurate because you have multiple different signals coming in and is robust because you can fill in the gaps in other ways.
So for us, many patients don't necessarily want to talk, they just want to play the video game, and they're completely fine to do that, and we will still gather data and be able to get insights on that particular day. The other thing is If you just use one modality, it makes it incredibly hard to expand to other disorders. Hence why a lot of these other companies just focus on one. Whereas for us, it's super easy to expand. And indeed, We already are. So although there are other companies, nobody is doing it in the same way as us. The other big thing about thymia is, we've addressed the whole assessment process as part of a much larger end-to-end solution. So we don't just go in and do a one-off assessment, like a lot of competitors do, we don't offer our service just as a single API; our service kicks in from the moment the patient first visit the clinician, and we follow them through the treatment, you know, aging, and empowering both patient and clinician every single step of the way, which means we become a much more integral part of the system, we offer a lot more value. And also, it means we get a lot more data from this end-to-end approach. And that's something unique; we are the only people to be doing that at the moment.
You know, one of the things we've seen as a result of COVID is that digital health is doing well as an industry. Do you think that people are more receptive to these types of tools as a result?
Absolutely Yeah, so I think COVID-19 hasn't itself, of course, brought on a lot of other issues. Many people have said a mental health tsunami with it, but it has also really accelerated acceptance across age groups of digital health. And it's open doors to new ways of dealing with these things. One of the most interesting things I've heard recently, well, not recently, so much anymore. But you know, maybe midway through the COVID crisis was universities, we were speaking with the mental health departments at universities. They said to us students are doing much better at the moment than usual. And when we dug in a bit more, it was because students felt a lot more comfortable interacting with their therapist or a therapy group through their phones. After all, this is what they're used to. And their response to treatment was much better as well as a result. So there's a very interesting, unique kind of new thing that COVID has brought with it. And I also think, for sure, we're seeing even 70-year-olds responding much better to this kind of digital solution. I think also there are government mandates at the moment that are pushing for digitization and mental health. And the NHS has a lot more accepting; I think the cogs would have turned a lot slower on this if COVID hadn't pushed hard, essentially.
What springs to mind - is this reimbursable?
So at the moment, we're not working with a reimbursement model, but certainly, it's in our plans to do so.
Alright, because that's typically a big
It's a big thing for the states as well. And Germany and other areas. So yes, it's, it's on our roadmap.
Looking forward to the next year? What do you think your big challenges are?
One of the biggest challenges we're facing now is hiring and quickly enough to meet the growing demand. And, of course, that's, you know, to do with funding, and it's to do with, you know, just how fast you can hire people we're looking for. So we're looking for technical people a lot. So we're looking to triple our technical team. So there, you're looking for machine learning engineers, back-end developers, and front-end developers looking for game designers to help us create more of the games. The illustration is very important, as well. But also a few people, obviously, on the commercial side, as we
ll. But mainly it's technical and scientific. So that's one of the things we were concerned about initially, that engagement and people coming back to the platform might be an issue, particularly for people with major depression. And if they don't, this is not going to work because it's a longitudinal model. But actually, we were amazingly, positively surprised by the reception. Over 95% of everyone interacting with the platform has enjoyed the experience. And over 85% of the users have said they would continue to play the video games just for fun, even without the output. And this was the same for 18-year-olds, as we have 75-year-olds, so those statistics are the same across all age groups, which was very positively surprised. So we thought, okay; maybe this is something that younger people will enjoy more of. Even 70-year-olds will enjoy it.
Yeah, we're seeing seniors embracing digital tools. Digital has been around for a while that you know, they get the benefits of it. So we see that. We very much appreciate your time today. Thank you so much for joining the conversation.
We look forward to great things for you and your business partner, so we look forward to chatting with you in the future.Â