#5 Up and to the right: How Limbic is tackling the supply-demand problem of mental health care
plus... Akili & Virtual Therapeutics merger, $12m fundraise for maternal mental health, a spotlight on Flow Neuroscience and all the latest news in mental health innovation
Hi friends,
If you’re new here, welcome!
I'm Steve Duke and this is where I write about what is happening in mental health innovation.
My goal? To give founders and operators actionable insights they can use to maximise their impact in mental health.A few weeks ago I wrote about Limbic, the UK based company developing mental health AI tech. They had just raised $14m in a round led by Khosla Ventures and the more I read about them, the more I was intrigued.
Their product had been adopted by 40% of NHS Talk Therapy providers with over 280,000 patients using it. They’d also published a bunch of impressive research supporting positive clinical outcomes.
I wanted to learn more.
What was the story behind this company, what were they getting right and what were they planning for the future?
So I caught up with Ross Harper, CEO and Co-Founder of Limbic for a chat.
In today’s post, we go deep on Limbic, sharing insights from Ross on what makes Limbic special and how they think about addressing our mental health crisis.This week in The Hemingway Report:
- Limbic: their approach to tackling the supply-demand problem of mental health care and what we can learn from them
- Maternal Mental Health: the startup that just raised $12.6m to expand virtual mental health services to women on their parenthood journey.
- Our Mental Health Funding Database: in case you missed it… we made a database of all mental health investors and grants which you can access for free.
- Flow Neuroscience: The company developing at-home brain stimulation headsets to reduce symptoms of depression.
Up and to the right: How Limbic is tackling the supply-demand problem of mental health care
“The hardest thing has been the fact that everybody told us we were crazy for so long.”
Ross Harper
Some people’s ovens just burn hotter.
Ross Harper is one of those guys.
Among other degrees, he has a PhD in Computational Neuroscience from UCL and when you speak to him, it’s clear he’s one of those people who just operates on another level.
He has an incredibly logical way of thinking about mental health both as a public health issue and as industry - one of the reasons I enjoyed talking to him so much!
I wanted to start by understanding exactly what problem Limbic is trying to solve.
There’s one chart that can sum it up….
Ross explains What’s going on here…
“There is an intractable workforce supply problem in behavioural healthcare.
There are simply not enough trained mental health professionals alive on the planet to deliver care to the number of people who require treatment. And we know that because we know what the disorder prevalence is in the population. So this is an economics issue. It's a fundamental mismatch of supply and demand.”
To make this economics issue worse, the trained professionals we do have, have to spend a lot of time on activities that are not actually delivering care.
According to a 2015 APA study, the average mental health professional works 46 hours per week in the US. Of this time, only about 50% of their time is spent on direct patient care.
So not only do we have a shortage of clinicians, but they only spend half their time on actual patient work.
The rest of their time is spent on admin, billing, client communications, teaching, supervision and more. Clinicians have to get paid for this time, but it’s ultimately not delivering care to the patient.
Oh, and if you aren’t aware, there is a serious burnout problem in the psychology profession due to the pressures caused by this supply-demand problem (we did some research on this last month which you can read here).
Vicious cycle…
Many people are aware of this supply-demand issue, but most organisations have taken different strategies to try and tackle it. Let’s try and understand how people are approaching this fundamental problem.
We can assess any mental health solution along two dimensions, (1) their ability to deliver clinical outcomes (x axis) and (2) their ability to scale (y axis).
It’s an interesting exercise to map different solutions onto this chart…
Human Based Psychotherapy is not a perfect treatment, but for many disorders, it is the best we have. Unfortunately (for the reasons outlined above) it is not very scalable.
We’ve got the “wellness” apps of the world, which can be delivered at great scale, but ultimately have less evidence that they actually improve outcomes for people, and are certainly not classed as clinical tools for use as part of treatment. Not much help…
There are many organisations trying to develop novel therapies like digital therapeutics or psychedelic assisted therapy, which have the potential to be both highly scalable and deliver great outcomes. However, it is still early stages for this category and there are significant differences in the robustness of the evidence base and the scalability of each of the treatments in this category.
Then we’ve got telehealth businesses. Organisations who are shifting the provision of talk therapy online. This does improve access for patients in more remote locations and has some marginal improvements on their ability to scale treatment (by allowing clinicians to work from anywhere), but it doesn’t move the needle significantly. This treatment is still reliant on a clinician spending one-on-one time with their patient.
Limbic takes a different approach.
They aren’t focused on developing their own treatment or another teletherapy service.
Instead they are focused on amplifying human-based talk therapy. Using an AI copilot, they are helping to increase the scalability of this treatment whilst also improving clinical outcomes.
They are shifting human based talk psychotherapy up and to the right, attempting to fill that important quadrant.
This approach comes from Ross’s core belief that human therapists are a critical part of the system.
“They are highly qualified, highly intuitive, and I don't think we should downplay their value in the system. I feel it would be a very arrogant thing for a tech company to believe that what they can build today is anything close to a substitute for the expertise delivered by a human therapist.
Secondly, I think that there is a very good case that therapy is inherently human, which makes it challenging to recreate with a one hundred percent technology play.”
And this is coming from a PhD in computational neuroscience who runs an AI company…
The research supports Ross’ views.
We know that about half of the therapeutic benefit of psychotherapy comes from the human relationship between the clinician and the patient, and I’m yet to see data that suggests other non-human forms of therapy can replace this.
North stars
Limbic have two guiding metrics for the work that they do – clinician time savings and patient outcomes.
I could write a whole post on defining the right metrics but what I like about these is that they are counterbalances to each other.
There’s a very easy way to save clinician time… cut therapy sessions to 25 minutes each. But guess what? That would obviously negatively impact patient outcomes.
These guiding metrics also create alignment with improving mental health at a systems level. Go back to our chart for a second. Saving clinician time ultimately improves scalability by alleviating some of the supply gap. And improving outcomes is obviously a good thing too.
By using these two metrics in balance, Limbic have a clear north star.
Up and to the right my friends!
So what does Limbic actually do to achieve this?
Limbic’s core product is Limbic Access, which is focused on improving the referral journey for patients. They call it a “digital front door” for self-referral, and I would say this is pretty accurate.
At its core, the product is an AI chatbot that providers install on their websites. Visitors to the site can chat with the bot and the bot helps them find the right care for their situation.
Without this, patients would have to call the provider (which can only happen during office hours and can be a pretty daunting experience for people new to treatment) or fill out a form that is honestly, disgusting…
Here’s one such form taken from a provider website….
And that’s only half of it….
But here’s what it looks like with Limbic…
While the patient is engaging with the chatbot, Limbic Access does some really interesting stuff for providers on the backend.
First, it collects a lot of referral information like contact details, demographics, eligibility criteria and communication preferences. All the stuff that would be collected in the big form, but in a much more engaging medium.
Limbic is connected with provider Patient Management Systems, so it can then pass this information directly into the relevant fields or notes for the provider.
The second main job it does is in screening and prioritising risk. If a referral is not eligible for that particular service, it’s able to refer them to other relevant NHS services through their integration with NHS Spine. This is super important as it avoids the user encountering the “COMPUTER SAYS NO” situation which we all hate!
Thirdly, Limbic uses specially trained statistical models to identify the patient’s primary presenting problem. Using the patient inputs that it has collected, combined with its clinically validated AI models, it computes the right ADSM to administer, and calculates the most likely presenting problem.
What are the benefits of this?
For the provider, it saves a bunch of time. Remember how Ross is super focused on saving time for clinicians? Well this is how he is doing it!
And for the patient, it is a much more engaging experience.
We must remember, seeking help is daunting. You don’t know what to expect and for many, it’s a big step.
So if we can hold the hand of the patient from the moment they choose to get help, making them feel comfortable and listened to, and directing them to the right care for their needs, that is a fantastic outcome and will be linked to higher engagement with treatment and ultimately better outcomes.
Limbic’s research shows the impact this has for patients.
It increases the rate of successful self-referrals (especially among minority groups), reduces changes in treatment pathway by 45% (great for both patients and providers) and in general, just seems to be really liked by patients (98.7% patient approval rate).
Limbic Access is their flagship product, it’s what has the most traction within the NHS and now in the US, but they have another interesting product that they launched earlier this year…
Limbic Care.
One of the biggest challenges in talk therapy is patient dropout. Research suggests that between 35-40% of patients drop out of therapy before completion of their treatment. That’s a lot. Sounds obvious, but if we want to improve clinical outcomes, we need to keep patients engaged with care.
This is what Limbic Care is attempting to do. It’s filling the gaps between therapy sessions by providing a range of activities in a chat-based, bite-sized format.
I like this approach.
During my own psychology sessions I have to scribble down my homework tasks in my notebook and then remember to work on them between sessions. Any material is emailed to me as a scan or PDF.
Not exactly 21st century stuff.
Limbic Care is different. It’s a well-designed app that is highly engaging for patients.
And the results are great. They’ve been able to decrease drop-outs by 30%, increase session attendance by 4% and improve clinical outcomes with less resources required by providers. This is pretty amazing.
There are a bunch of activities clinicians can assign as homework, it has a chat feature and of course, crisis detection functionality.
Now I’ve seen a few companies playing in this space. But one problem they tend to run into is user engagement. Patients just stop coming back to the app.
So what’s the difference between these and Limbic Care?
It could be the use of the chat based interface. Yes, Limbic Care has a chatbot that users can interact with in the app for general queries, but the main difference is that the homework activities themselves are largely delivered using the chat interface.
Now, if you’ve tried any of the generic mental health chatbots from the app stores, then you know just how clunky they can be. It’s like having a conversation with a vending machine.
Due to recent advances in language modelling, the bar is now much higher for what users will actually interact with.
Limbic is one of the few companies who’ve taken these advantages in large language modelling and brought them into clinically safe practice while still meeting users’ expectations of quality.
And that might just be the key.
It’s a medium that people are highly comfortable with and allows for much more tailored and engaging content, as opposed to static slides.
Currently, Limbic Care is focused on patients engaged in group therapy, trying to reduce drop-out and improve clinical outcomes, all without the need to increase time from clinicians.
Up and to the right…
What can we learn from Limbic?
Limbic is making the kind of progress that most early Mental Health businesses are aiming for. So what can we learn from them? What are they doing that’s allowed them to achieve this success?
1. They’re building with the system
Health systems are clunky beasts. We know this.
They’re full of the kind of red-tape, archaic systems and bureaucracy that deters the rebellious spirit of many founders.
But once you understand the mental health landscape, you realise that whether you like it or not, the public mental health system is at the core of providing patient care.
They are the largest payor in every major country and if you want to solve this problem for the entire population (not just the wealthy people who can afford private health), then you need to play ball with the health systems.
Limbic gets this.
The NHS and providers are all focused on increasing access to care and waitlists are a real problem for them. Ross and team are aware of this and have built solutions to directly address these problems. They are aligning themselves with the priorities of the system that they are part of (and ultimately, who pay for their products).
From day one, they also had a big focus on building for providers.
The reality is that without having healthcare providers on board, it’s very difficult to get your product adopted by patients and paid for by payors. Limbic is data driven (25% of their employees have PHDs…) and Ross saw focusing on providers as the easiest way to build an evidence base for their solutions.
“As an evidence-based organisation, we wanted to show impact, and the fastest way to show impact was to be adopted by the people on the front line, solving their specific problems.
And I think it's a relatively unique business model, that we focus so strongly on that particular piece of the puzzle. We don't go directly to consumers. We haven't tried to move far up the chain to these gigantic payer organisations. We focus on solving problems for providers and amplifying them to reach more patients.”
They also understand that people are sceptical of AI.
It can be frustrating for founders to see their product not getting adopted because of misplaced hesitations about certain technologies. But guess what?
Standing there screaming “JUST BECAUSE YOU DON'T LIKE THE SOUND OF IT DOESN’T MEAN IT DOESN'T WORK!” won’t actually convince anyone to adopt your product.
Ross recognised this from early on. That’s why Limbic approaches providers with a partnership mindset.
“From the beginning, we sort of had this partnership model with the care provider organisations we were working with, where we had to educate them on how the AI worked, help them understand different things they may be looking at in the market, help them understand where they should focus, where they should not focus, and teach them how to do this safely.
And it's been a really beautiful synergistic relationship where they're educating us all the time and we're improving our clinician training program as we do our best to help demystify this technology.”
2. They’re not afraid of hard problems
Limbic is building for providers, so I asked Ross, how do you balance this with having to also build for patients? Which one do you prioritise?
His answer surprised me but I loved it.
He told me that they don’t prioritise. They take it on as a dual challenge.
I’ve had this conversation with multiple startups in this space – “who should we prioritise on our roadmap?”. The hard, but correct answer, is that we can’t prioritise one group over the other. To create an impactful solution, you need providers AND patients to love what you’ve built.
“You need to understand the specific pain points that you're solving for within each user group and build specifically for them. But I'm not going to shy away from the fact that it's a double product challenge.”
Of course, building for two different parties is much harder than one. But that’s what I like about Limbic. They’re not afraid of hard.
There’s a great lesson here.
At Wayflyer, when I was looking at different markets we could enter, I’d initially rank them by “ease of entry”. If it was a favourable regulatory environment and we could get in easily, I’d suggest we prioritise the country.
But Aidan Corbett (Wayflyer CEO) taught me an important lesson.
He told me that every other company will do the same analysis and reach the same conclusion. He told me that no-one wants to do the hard thing. So let’s go do that. Let’s do the work to get into the countries that are hard to enter, precisely because it’s hard and that will deter all the companies who don’t want to do the work.
3. They’re nailing the basics
Michelin star restaurants don’t just have great food, they get all the basics right, all the time.
They get the lighting and music just right, they know exactly how to greet you and take your coat on arrival, they even fold your napkin and leave it on your seat when you go to the bathroom.
This focus on getting the basics right is a common trait in high performing organisations and I see evidence of this at Limbic.
Not only have they built great technology, but they’ve done the work to translate it into multiple languages. They’ve built Limbic Layer to address patient safety concerns. Even the blog on their site is designed well and written with clarity.
Success leaves clues.
They aren’t just cooking great food, they’re folding the napkins too.
Now, a restaurant without great food will never succeed.
But if you’re nailing the food, getting all the other small stuff right is what helps you stand out above the competition. And in Limbic’s case, I’m sure it helps them to win the kinds of contracts they’ve won with the NHS and others. Public Health Systems care about the details!
So what does the future hold for Limbic?
There’s still 60% of Talk Therapy providers in the NHS that Ross wants Limbic to be adopted by. That’s a core focus.
On top of that, they’ve recently expanded to the US which I’m delighted to see. Of course, this market poses its own set of challenges that Limbic will need to navigate (not least, an entirely different payor landscape).
“The vision is this. We want to make the highest quality mental health care available to everyone everywhere, regardless of their socioeconomic background. And in order to achieve the vision, we can't stay in the UK.
So we have to grow, we have to take this to the masses, to everyone. So we are excited to be working with US payers and providers.”
Limbic Care is still a young product with the potential to improve engagement with treatment and outcomes for patients and I’m excited to see what they do. Although I didn’t get too many details, Limbic has a bunch of new AI developments in the works.
And in terms of the long term strategy?
“The boring answer is it will be an extension of what we're doing today. To me that's true scale: doing what you’re doing today, but just bigger.
It will be an AI copilot used by care providers and patients, and it will tip the balance for supply and demand in favour of patient outcomes, and it should bend the cost curve in behavioural healthcare as an industry more broadly.”
This has echoes of interviews I’ve listened to with Patrick and John Collision in the early days of Stripe. They would just talk about how many relatively mundane problems that still needed to be fixed in the payments industry.
There isn’t always a super sexy product roadmap. Often, the best companies realise that they just need to execute on their core strategy, continue to solve the most important problems for their users, expand into the biggest markets and try to achieve scale.
Bezos did this with Amazon. He knew people wanted products at great prices and delivered fast. So he spent literal decades building the infrastructure to do this.
It’s the kind of strategy that Pear Therapeutics could have done with….
It’s clear that with the current set of mental health treatments, mental healthcare has a significant supply-demand problem.
Reducing demand seems unlikely in the short term, so we have to improve the supply of care, both in quantity and quality.
This is what Limbic are doing, using technology to save clinicians time and give them leverage. And for patients, they are improving outcomes by making it easier for them to get into and stay in care.
They have deep technical capabilities, are well-funded and appear to have the kind of mission-driven orientation that makes me hopeful about the future of mental health innovation.
Mental Health Funding Database
In case you missed last week’s post, we built a database of mental health investors and grants available for mental health organisations. Check it out.
Hopefully this saves you a few hours of googling!
Mental Health News
Here’s your roundup of the top news in mental health this week;
- Virtual Therapeutics and Akili announce a merger. This is big news for anyone interested in digital therapeutics. (BusinessWire)
- Seven Starling, a maternal mental healthcare startup, raised $12.6m in funding through a combination of equity, shares and warrants, and is seeking an additional $1.7m to complete the round (BHBusiness)
- Irish startup Spectrum.Life, which delivers comprehensive digital health, mental health and wellbeing solutions, raised €17m in a recent funding round led by Act Venture Capital (EU-Startups).
- Victorian Government invests $6.4m into Orygen for their youth digital mental health service MOST (Orygen)
- Workplace wellbeing provider Habit Health acquired by Five V Capital for over $200m (AFR)
- Venture Capital investors seek out business models with proven track record and novel approaches, with increasing interest in enablement services (BHBusiness)
- Nasdaq-listed Acadia Healthcare creates new COO position for Dr. Nasser Khan to expand their comprehensive treatment centres, grow their outpatient segment and improve quality of clinical care. (BHBusiness)
Company Spotlight: Flow Neuroscience
- What they do: Flow Neuroscience develops at-home brain stimulation headsets that reduces symptoms of depression by sending gentle electrical impulses to the brain.
- Headquarters: Malmö, Sweden
- Founders: Daniel Månsson and Erik Rehn
- Founded: 2016
The Flow headset device aims to stimulate the frontal lobe through the use of transcranial Direct Current Stimulation (tCDS) technology. Flow devices send gentle electrical impulses to areas of the brain that have slowed down to reduce symptoms of depression.
tCDS has been around for a long time, with over 20 RCTs in the literature and several meta-analyses concluding that there is class A evidence for the ability of tDCS to improve depression. However, access to this technology was previously costly and generally only used at clinics.
Flow Neuroscience is enabling people to use the device easily at home with the assistance of a mobile application with users able to buy the device for €459 or rent it for €89 per month.
It is available through the NHS, UK’s Well Pharmacy and through their online store.
Make it this far? Fair play! Reply to this email and let me know what you thought.
That’s all for this week.
Keep fighting the good fight!
Steve Duke
Founder of The Hemingway Group
P.S. feel free to connect with me on LinkedIn
