— Could you please tell me a couple of words about how you met Dr. Stefano Goria?
— Stefano and I met at a startup accelerator called Entrepreneur First (EF). We were both part of the LD14 cohort based in London. The great thing about EF is that it brings together people with completely different backgrounds who may have complementary skill sets and similar interests and helps them to set up deep tech start-ups. I had the idea for Thymia before joining the accelerator, but it wasn’t until I met Stefano that the idea became fully formed with all of its working parts. As soon as we started talking we decided to partner up and work together to develop Thymia. We are both very mission-driven founders and it wasn’t just our skill sets that complemented each other but also our characters.
— You launched Thymia into a pandemic. How did it help the project?
— When we started working on Thymia in April 2020, it was against the backdrop of the pandemic, when the UK was also experiencing the beginnings of an intensifying mental health emergency. COVID-19 put an even greater strain on an already overburdened system, unleashing a new wave of mental illness borne out of social isolation, employment loss, bereavement, grief – all made worse by how hard it was for people to seek support. From January to March 2021, the ONS found that 21% of UK adults reported depressive symptoms – that’s twice as many as before the pandemic.
This context made it even clearer that new approaches to mental health assessment and treatment were desperately needed. It spurred us to transform the way we assess mental illness, by empowering clinicians with new digital tools that give them the objective data they need to have a better and more continuous overview of their patients, allowing them to make faster and more accurate clinical decisions. Our platform is an end-to-end solution that not only provides clinicians with objective data to ensure assessments are as accurate as possible, it also streamlines several processes before and after appointments. This saves clinicians time so that they can do what they do best, interacting with and understanding their patients.
— How much time did you spend developing Thymia? Is the latest version different from the first?
— So far we’ve been working on Thymia for a little over a year. The platform has been through many iterations already and we now have a product that is nearly ready to launch. The latest version is different to our initial prototypes in that it has been trialled and tested on a number of users, taking into consideration both patient and clinician feedback in the design, look and feel to maximise engagement. Aside from the user experience, a key element of how Thymia works is based on ethical artificial intelligence, where training our algorithm is an ongoing process that can never truly be classified as ‘finished’. Our algorithms are necessarily based on millions of data points and they become better as we collect more of these; it is imperative that we continue to gather data from a diverse range of users and improve our models so that we can ensure no bias is introduced that may be related to language, accent, ethnicity or gender. What’s more, although we are focusing on major depression now, as we grow to tackle other cognitive disorders our models will soon include data from other patient populations as well, such as patients with Alzheimer’s Disease or Parkinson’s Disease.
— How many people are currently testing Thymia? How are you going to monetize the platform?
— We are trialling the technology with over 2,000 patients and healthy individuals. Thymia will be completely free for patients to use but will be contracted by health providers such as private individual clinicians, mental health clinics, telehealth companies and NHS Trusts on a pay per patient per month basis.
— What kind of video games (a couple of examples) do you use to identify signs of depression?
— Our video game-style activities are designed to elicit speech and to target specific cognitive deficits typically associated with depression. Some examples include verbally describing animated scenes (such as an outdoor street scene featuring a cafe and various visual stimulants) and interacting with and memorising moving objects (such as cartoon bumble bees). Whilst users complete the games, the Thymia software anonymously analyses three key data streams: voice (how someone speaks and what they are saying, to pick up acoustic and linguistic depressive cues), video (micro-expressions and eye-gaze, which can help track current mood) and behavioural measures, including reaction times and error rates (which can help detect the core cognitive and physiological symptoms of depression). The software then identifies data patterns indicative of depression to help clinicians pinpoint a diagnosis more quickly and accurately, and also to work out whether treatments are working over time.
— In an interview with Sifted.eu, you talked about how the current system for diagnosing depression has many flaws. “This has nothing to do with the clinicians themselves, it’s really to do with the tools that they have”, — said you. What are the flaws of the current system for diagnosing depression? How does Thymia solve this problem?
— Currently, GPs do not have the time to screen for depression routinely; someone needs to first complain of feeling low in order to be assessed. Furthermore, in order to assess depression, GPs and psychiatrists rely on questionnaires that have been around since the 1960’s. Not only are these questionnaires subjective, easy to influence and prone to bias, they are also unable to capture the subtle nuances of the early signs of depression or track the complexities of ongoing mental illness. On top of this, there is no routine way to monitor patients between clinical sessions, at home, where most depressive episodes and suicide attempts occur. As a result, clinicians unfortunately end up missing a significant part of the big picture. When you combine these things, it means that identifying the right diagnosis and treatment can take years of trial and error. On the patient’s side, this system also means patients often won’t have a good understanding of their own mental health or treatment progress, making adherence to treatment poor.
We’re solving this problem by building a system that will supplement the subjective questionnaire-based approach to diagnosing complex mental illness and in time replace it. Our platform will step in from the moment the patient first visits a clinician all the way through to the end of their treatment, helping clinicians to objectively and continuously assess and monitor depressive symptoms. By using video games powered by Neuropsychology, Linguistics and affective Machine Learning on a daily or weekly basis, we empower clinicians with a more comprehensive picture of their patient’s mental health. This helps remove the subjectivity from mental health assessments and speeds up assessment and treatment: instead of answering questionnaires, patients play interactive games which detect signs of depression, as well as monitor whether symptoms are improving or worsening over time.
— Do you have competitors in the UK?
— We don’t currently have any direct competitors in the UK; we are the only company to use multiple data streams to assess mental illness. Our philosophy is that mental illness is as real and objectively measurable as physical illness and we strive to show this. In doing so, we hope to eradicate the stigma associated with mental illness and encourage people to discuss it more openly and inclusively, as well as helping clinicians to deliver the best treatment to their patients. Our technology will empower clinicians to assess and treat depression faster and more accurately, whilst allowing patients to develop a deeper understanding of their own condition.
— What are your plans for the further development of the project?
— We’ve already established collaborations and partnerships with several world-leading research institutes, including UCL and King’s College London, to run further trials and help scale use of the technology to other cognitive disorders. We are already looking to apply our technology to help assess and monitor other conditions including Alzheimer’s Disease, Lewy Body Dementia, Parkinson’s Disease, Anxiety, ADHD and Autism.
— What markets are you interested in now and why? Is Russia on your list?
— We are starting out in the UK, but we already have strong interest from various countries in Europe and South America. Our next expansion targets are Norway, Germany and Brazil. We are certainly interested in Russia as well; our intention is to become the gold standard of cognitive assessment globally.
By Olga Grinevich
Photos provided by Thymia