At Luminous, mental health is a big theme for us and we have made three investments in the area in recent years — BioBeatsOxford VR and Ellipsis Health. However, one persistent question is the ability of apps to alleviate anxiety or depression. Based on current learnings, the answer is yes — but the catch is that people struggling with these conditions can find it hard to engage with the very tools that might help them. You might call this the “engagement paradox”. But new research is improving our understanding of how we can make digital mental health tools more accessible and affordable for those who need them most.

Cognitive-behavioural therapy (CBT) is widely used to address a range of mental health issues. A talking therapy which aims to help people change the way they think and behave, it is considered to be effective in tackling anxiety and depression, as well helpful for numerous other disorders [1]. But therapists may not be available or affordable, while would-be users may be prevented by disability from attending in-person sessions, or unable to fit them in around work or domestic commitments. The Covid-19 pandemic has further exacerbated the need for remote mental health support.

Over the past decade, therefore, we’ve seen the rise of internet-based cognitive behavioural therapy programmes (iCBT): app-based videos, audio guides, quizzes and activities which can range from techniques for improved breathing to ways to recognise distorted thought patterns. Most clinically recommended iCBT is done in conjunction with a remote therapist who can monitor outcomes and select topics to address specific needs.

Does iCBT work? It seems to. Several clinical trials have shown that iCBT can deliver significant improvements, with one recent study involving more than 360 patients [2] reporting that after three months more than half no longer had a diagnosis of anxiety or depression. But other studies have returned comparatively disappointing results [3], finding that iCBT added little extra clinical value compared to a routine consultation with a GP. Why should there be such a disparity in outcomes?

Degrees of engagement
One major reason seems to be differing levels of engagement between patients and platforms. Several studies have shown that younger people and those whose symptoms are more severe are more likely to drop-out before completing the full programme — the latter being particularly worrying, since it means that the patients who could benefit most from iCBT are also the ones most likely to struggle to engage with it. This frustrating situation urgently needs remedying if the benefits of iCBT are to be fully unlocked.

Earlier this year, a collaboration between Microsoft, Trinity College Dublin and SilverCloud Health initiated one of the largest ever studies of how patients engage with an iCBT platform. Published in JAMA last month [4], it surveyed how more than 54,000 users interacted with a 14-week programme. The researchers used machine learning to group users into five distinct categories based on the time they spent using the platform as a whole, and on each of the different types of therapies offered within it — which included textual and visual guidance on subjects such as sleeping and relaxation, and mood trackers, quizzes and other evidence-based CBT exercises. All users received guidance from therapists for the first eight weeks.

The five categories included low engagers (36.5%), who used the platform very little; late engagers (21.4%) who showed lukewarm, but consistent interest, initial high engagers (25.5%) who started out strong but whose use subsequently plummeted; high engagers (6%), who showed a more modest decline in use; and the highest engagers (10.6%), who completed all the modules. While all five groups experienced some level of benefit, the highest engagers unsurprisingly demonstrated the most consistent clinical improvement — but were one of the smallest groups.

Dimensions of engagement
The study offers support for the use of iCBT, but perhaps more valuably offers some pointers on how to improve overall outcomes. For example, the study found that the low engagers were more likely to use mood monitoring tools, while the highest engagers favoured journaling. Developers could use such insights to create an engagement profile from a user’s behaviour and adapt the programme accordingly. That should help more people to complete more of the programme and hopefully reap more benefit.

This is only a beginning. Mental health conditions are complex and “length of time spent” is a very blunt metric, as pointed out in an editorial accompanying the JAMA study [5]. For patients experiencing severe depression or anxiety, for example, simply logging into an app can feel like a major accomplishment. That would not show up as much time spent, but could be helpful to their recovery. Other studies have suggested that a person’s willingness to return to a programme could be more telling than time spent. Future studies could explore this in more detail, by taking into account patients’ starting points when evaluating their progress and assessing other dimensions of engagement.

As we have seen from our experience with Oxford VR, CBT delivered via virtual reality (VR) has the potential to be much more engaging than internet delivered treatments. First, VR should, for many conditions, be more powerful than internet-delivered ones on the basis that it delivers the key experiential learning that leads to change, and it is less passive than internet delivered treatments. Second, VR can deliver treatment in imaginative ways that take over the main senses (for example, a patient is less likely to get distracted with a headset on). Longer term, VR also the potential to collect more real-life equivalent data (e.g. where person looks, where moving etc, physiological markers) which could be used to tailor treatment more effectively.

Ellipsis Health uses AI to detect indicators of heightened anxiety or depression in speech patterns — the kind of monitoring that will be key both to improving engagement and ensuring further help is available if needed. We thus look forward to further steps towards resolving the engagement paradox, in the service of better mental health for all and we are excited about working with entrepreneurs to improve patient outcomes in this critically important area.

[1] https://www.frontiersin.org/articles/10.3389/fpsyt.2018.00004/full

[2] https://www.nature.com/articles/s41746-020-0293-8

[3] https://www.bmj.com/content/351/bmj.h5627

[4] https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2768347

[5] https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2768343