Health Tech World hears from the co-founder and CEO of Thymia, a UK startup that is using AI and biomarkers to bring objectivity into mental health assessment.
According to some studies, fewer than 50 per cent of GPs cannot accurately identify a mental health condition, while 75 per cent who experience a mental health condition will never get the help they need.
As waiting lists soar, neuroscientist Dr Emilia Molimpakis was inspired to build better tools for mental health care after seeing her best friend struggle to get the help she needed.
Suffering from depression while at university, doctors and a psychiatrist failed to spot the severity of her condition.
Dr Molimpakis saw a system based on highly subjective data, collected during short appointments with questionnaires that do not always work.
So, she launched Thymia, a digital tool that aims to make mental health assessments more objective and empirical.
The tool uses AI-powered video games to analyse millions of biomarkers, including patients’ voices, facial expressions, movements and behaviour.
The AI models have been developed using Thymia’s proprietary data set of over 1 billion data points collected from over 5000 individuals and over 200,000 game sessions.
The company claims its models are the only ones globally that combine voice, video and behaviour data in this way.
The platform is being used for non-clinical mental wellness assessments, but a recent injection of funding will be used to scale the technology in the clinical space through medical device approval.
Last month, Thymia announced a $2.7 million seed round to scale the technology in the clinical space and build out the AI infrastructure.
Currently, the tool presents up to 84 per cent accuracy in detecting clinical depression and anxiety.
While the app is currently operating as an investigational device, it has the potential to dramatically reduce the time to diagnosis for mental health problems, helping clinicians identify the right intervention sooner and providing insights during the gaps between appointments.
As the company continues to raise investment through a Crowdcube community fundraise, Health Tech World caught up with Dr Molimpakis to hear about the personal experiences that inspired the company, the technology that drives it and the future of tech-enabled mental health care.
Health Tech World: Thymia was born out of a personal experience. Could you tell us about your friend’s experience with mental health services?
Dr Emilia Molimpakis: My friend and I were both academics at UCL so we spent a lot of time together.
She was my best friend there throughout my master’s and PhD.
She very much went through something that a lot of academics go through, which is developing feelings of anxiety.
I think she was always quite an anxious person but over time, especially during the PhD, she started to develop stronger signs of depression.
She initially tried to get some help from the NHS. But she had to wait for months to see a mental health professional.
During that time, she deteriorated quite a lot. Eventually, instead of waiting for nine months, she paid to go privately to see a therapist.
They said she should see a psychiatrist and unfortunately, he also didn’t realise how bad her condition was.
He went through what psychiatrists usually go through, which is a bunch of questions which are quite subjective.
In the end, the psychiatrist got a very different view of the situation than what was actually happening.
Just two days later, she was given a prescription for antidepressants. But sadly, she tried to take her own life. I was the one who found her at her house.
It was very traumatic. I consider it very unfortunate but also very fortunate, because it could have ended differently.
She is stable now. She’s getting help. But the situation really pushed me into seeing how problematic this system is and how many gaps there are.
Can you talk more about these gaps in care that led to your friend’s condition not being recognised?
I’ve spoken to her psychiatrist and a number of others. Ultimately, they all said the same thing.
At the GP level, [they] don’t have time. At the specialist level, [they] have a bit more time, but don’t have the right tools.
They have questions and observations, and the patient can very easily mask what is going on.
That happens all the time. And for depression and anxiety, it means you can go 10 years without getting a proper diagnosis or the right treatment.
On average, it takes over 10 years to get the right treatment.
In that time, you deteriorate; there can be self-harm episodes and suicides. All things that could be prevented, but aren’t.
You launched Thymia at the very beginning of the pandemic. How has the mental health landscape changed in that time?
The pandemic had a really amazing impact on people’s acceptance of telehealth, which makes things a lot easier.
Prior to COVID, about 90 per cent of all mental health appointments were being done in person. Now, it’s the opposite.
The majority of people prefer to do things online because it’s a lot easier. It’s private, you feel a lot more comfortable in your own home discussing mental health.
The landscape hasn’t changed in the sense that there just are not enough specialist mental health clinicians and they still don’t have the right tools.
COVID has more than doubled the rates of depression and anxiety symptoms. So it has opened up a whole amount of telehealth, but at the same time, it’s made the situation worse.
Part of Thymia‘s mission is to bring objectivity into mental health assessment and monitoring. Can you explain how the technology enables this?
Thymia brings my background in neuroscience and linguistics together with our chief science officer’s and chief technology officer’s backgrounds [in] AI.
We analyse three different types of data streams to look for signs of depression, anxiety and other conditions.
Specifically, we look at voice. We split that into acoustics and content, so how you sound and what you say.
Secondly, we look at video. This is everything we can pick up from a smart device camera.
Originally, it was eye gaze patterns and facial micro-expressions. Now, we’ve expanded to upper body movements like twitching and the angle of your head.
The last category is behaviour. This is how you’re typing on your phone, swiping and reacting.
We gather all of these data through specially designed video games. This is how we engage with patients.
How do clinicians engage with the tool?
We’ve been able to not just look at whether someone has a mental health condition, but more importantly, we’re the first company that we know of to look at the symptoms of those conditions.
This is really important because ultimately, clinicians don’t treat the condition, they treat the symptoms of the condition.
We can look at things like fatigue, memory issues, mood swings or loss of interest in doing things.
We will tell a clinician, ‘These are the symptoms, these are the levels of each one, it’s down to you, as a clinician to decide how you think this person should be treated and how you think they should be diagnosed’.
We’re operating like a tool rather than a diagnostic. It can’t replace the clinicians’ opinion.
Does the platform account for nuances related to factors like culture, language and age differences?
Thymia factors in things like socio-economic background, language, age group, gender and gender identity. All of these things have an influence on the core data points that we’re looking at.
Your voice is affected by whether you’re born male or female. It will influence the pitch .Your age will influence whether your voice is breaking, for instance.
We gather hundreds of variables on everyone.
We ask questions so that when you come as a new user, we can use that information and say, ‘Okay, you are this gender, this age, et cetera. Let’s find people who match you so that we can isolate what is depression versus what is your age.’
On a wider scale, what role do you see tech playing in mental health care moving forward?
I think we are on the precipice of a massive revolution in mental health.
If you look back over the last 100 years ago, we had a revolution in physical health which was brought about by new tools like blood pressure cuffs, blood tests, thermometers and X-rays.
They really changed how physical health is monitored.
You couldn’t imagine a situation now where somebody was diagnosed as being diabetic without having done blood tests, urine tests, etc.
I believe that this is what we’re going to see in mental health thanks to our ability to look at the behavioural biomarkers that are just now being discovered.
I really do see the landscape changing, both in the acceptance of [mental health] and in the biomarkers being used, not just for treatment, but for developing treatments as well.
If they use these biomarkers in pharma trials, they can save a lot of money. They can also get to the right dosages and treatments a lot faster.
It’s the same with psychedelics. We’re not just looking at the traditional treatments, we’re also looking at novel treatments. Tech is core to making all of that work.
Things have changed a lot since the pandemic. But we’re going to see a massive change in the next 10 to 15 years.
Healthcare innovators and leaders honoured at Imprivata HealthCon user group and awards ceremony
Wearable communication system may reduce digital health divide
Molecule trains the immune system to prevent cancer
Diabetes patients urged to use fitness games with caution
Urgent work needed to tackle ‘substantial’ digital health inequality
eSight: “The technology has the potential to change someone’s life”
Microsoft invests £2.5 billion in UK AI
TMS shows promise in tackling depression ‘epidemic’
AI depression app set for NHS clinical trial
UK Biobank releases world’s largest single set of sequencing data
- Diagnostics4 weeks ago
3D model will advance understanding of spinal injury pathology
- Opinion1 week ago
Why it’s time to revisit workplace mental health initiatives and make them work for everyone
- AI3 weeks ago
AI can help predict survival outcomes for cancer patients
- Research5 days ago
Tackling the diagnostic testing sustainability problem