It's great to hear about your experiences with CBT and hear from someone in the ~50% for whom CBT works. We certainly wouldn't bill it as a panacea for all conditions, as we've discussed in the comments section here, and it's wrong to see it as such, but for those who it works for it's an excellent tool. Also great to hear that so-called bibliotherapy has been of assistance to you.
As you allude to, while there are many academic papers showing the efficacy of CBT, DBT and ACT, in various forms of trial, some other forms of talk therapy are not backed by such a robust evidence base.
Thanks for bringing those up. We're certainly aware of Quirk and the story, although we haven't as yet talked to the founders in person. Their story is very relevant to this discussion.
The points about sustainability of business model are very relevant ones, which is why we have a long term aim of building software as a medical device, which is potentially distributable / reimbursable by healthcare systems / insurance. This also opens up potentially different payment models, which can be more sustainable.
We've most certainly heard of Woebot. Would be really interesting to hear how you feel about Iona in comparison.
Obviously we're much earlier stage than Woebot who have in fact been around since 2017, but the mission is certainly similar. We've gone for a different tone and feel to Woe at the moment and will continue to expand and improve the product and content as the company grows. I think there's a demographic for whom the tone of Woebot doesn't really appeal, which is why we've gone for a slightly different style in our content.
We're generally more in the smart scripts camp at present. The conversations are graph based, with each node being associated with code that can look at the state of the user and any input. It may be that the input is a set of fixed options, or that the input is freetext. Usually in the event of freetext we're classifying that freetext, looking at the state and using these two things to determine the output and next node to go to.
It's still early days with the product and to date we've generally avoided anything that is generative, as you say, it's hard to keep on track and is hard to predict what the range of outputs can be. The next steps in development will most likely be to improve classification of freetext and use it more extensively within the app. We've also considered having individual nodes which use generative output (at least in part for the output), to go more to the generative end of the spectrum, but retain control of the conversation flow, however this is still at the concept stage for the moment.
We've seen some graph based / smart script engines where you can enter subgraphs where the output is essentially seq2seq until various trigger conditions are met (e.g. the text is classified a certain way), before continuing to another part of the graph, and this is another way the two approaches can be blended. It's not something we're looking at at present, and can also be potentially quite unpredictable.
"Generate chit-chat until the intent classifier gets what you need" is an interesting idea. Should work especially well in therapy where asking the same question for different angles is pretty well tolerated.
I agree, CBT is certainly no panacea, and as mentioned in the post, for those at the moderate to severe end of the spectrum, or for those where CBT is ineffective, it may be the case that medication is the only viable solution.
Y Combinator has actually funded several companies looking at the usage of various drug-based mental health treatments, including Osmind (S20) and another company in this batch (S21) who are looking at the use of psychedelic therapeutics for mental health, although none are addressing your specific issue with availability / accessibility which is a tricky one with respect to legislation and regulation.
Great to hear about your positive experience with CBT. I think there's a lot of innovation that's possible in this sector. If we look at CBT, a lot of the core concepts and exercises haven't changed all that much (from the perspective of the average spectator) in the last 20 or so years, which leaves a lot of potential research opportunity that can be explored using new technologies which weren't available when CBT emerged.
In fact, privacy is a concern that sometimes even puts employees off using their employer's Employee Assistance Program (EAP), even though that's often in the form of a crisis telephone number which you phone.
There are a number of apps in this space who sell to employers but there's no industry wide survey of utilization numbers for such apps, so it's difficult to judge to what degree this is an issue across the entire sector. We do what we can, offering completely anonymous sign up, no requirements to enter name, demographics etc., no location tracking etc. and it's certainly an important issue for users which we're cognizant of
Thanks and some great points here. As you say competition and differentiation are potentially really tricky in this sector. Really interesting to hear about your experience at ginger.io .
We're certainly not wedded to the chatbot idea, but it's just the best UX/UI we've implemented so far. The key we feel is to avoid becoming a boring, skippable content board, and maybe a chatbot isn't the final iteration of that concept, but it's the approach that has worked the best for us so far.
I agree that the key in this space will be to learn quickly and iterate to find a solution compelling to the user and payer. It's early stage for us, until the point we got into YC we had taken no funding, so I think there's a lot of scope for how we can expand and adapt our offering to the marketplace and we've already learned a tremendous amount.
We have ultimate long term aim of building apps which are certified Software as a Medical Device and get device-like reimbursement through healthcare systems. We certainly feel that's a place which is less competitive and more differentiated, there are a few companies in this space such as Pear Therapeutics, but unlike them, we feel it's important to start at the b2b and b2c end of the spectrum to build a product that people love before going through the lengthy and costly process of adapting that product becoming a fully certified medical device.
Thanks for bringing this point up, and it's a point which is often cited in favour of greater regulation of the sector in general - do such apps in fact stop people seeking help when they may have otherwise done so? Similar criticisms can be levelled at Calm and Headspace - did people not seek advise from a clinician when they actually needed it because they had Headspace? Could the existence of such apps be worse than nothing? As an emergent field of research there is not yet much definitive evidence on this subject.
It is also worth noting that this is not a new phenomena. The same could be true of people buying CBT textbooks, like the famous "Thinking good, feeling better" and other titles from the same author Paul Stallard, which have been around for many years. Maybe some people who bought that book got nothing out of it and were put off seeking therapy because they read it. There is little definitive hard data on the subject, but we do know that both app-based CBT and bibliotheraphy can both be effective (citations in original post).
If a user churns, there are many confounding variable associated with the loss of that user. You do not know the exact reason why they stopped using the app, and whether the impression of the app influenced their overall view of CBT in a positive or negative manner. Additionally, since durable recovery rates from in-person CBT are around 50% (citation elsewhere in this comments thread), you do not know if that user was in the category who would've benefited from CBT in the first place. They may have got a similar impression had they seen an in-person CBT therapist. Short of capturing the user post-churn and asking them in person, it is difficult to get a firm handle on such figures, but it's certainly an interesting and important statistic which hopefully will become clearer as research on the area progresses.
As you allude to, while there are many academic papers showing the efficacy of CBT, DBT and ACT, in various forms of trial, some other forms of talk therapy are not backed by such a robust evidence base.