Big data can help us take digital mental health interventions to the next level. That was one of the key messages delivered by Dr Jorge Palacios, Senor Digital Health Scientist at SilverCloud Health, when he spoke at the recent DTx Europe https://www.dtxeurope.com virtual conference.
At a time when interest in digital healthcare has never been higher – partly driven by demand, access and logistical challenges highlighted by the COVID pandemic – the ability to demonstrate effectiveness of online cognitive behavioural therapy and the online CBT app provided by SilverCloud is vital.
By presenting work which has leveraged SilverCloud’s access to big data, including millions of unique interactions from hundreds of thousands of users, Dr. Palacios was able to show not only that online CBT works, but that it provides better results than guided self-help bibliotherapy and psychoeducational group therapy, other low-intensity interventions which are similar in clinical content. He also demonstrated how SilverCloud is able to use the data analysis and research it conducts to provide valuable feedback to users’ supporters, helping to tailor their training, support their clinical decision-making and ultimately to deliver improved patient outcomes.
In keeping with SilverCloud’s vision, Dr Palacios based his conclusions on robust, rigorous clinical and academic research – in this presentation he explored three key areas that drew on real world data, not data taken from a controlled, trial environment. These areas included:
- Clinical outcomes
- User engagement
- Supporter messaging
Clinical outcomes
SilverCloud is deployed as a low intensity intervention for conditions such as depression and anxiety. The big data research Dr Palacios presented compared digital interventions (online CBT) with two other interventions: guided self-help bibliotherapy (GSH) and psychoeducational group therapy (PGT).
The data gathered included detailed information from 20k+ SilverCloud users from over four years which provided clear results. It used propensity score modelling to show that the three interventions worked – all producing improvements for patients. But critically, it showed the biggest improvement was found with online cognitive behavioural therapy.
Engagement
A further, larger data set, drawing on over 54k SilverCloud users, was used to examine different types of user engagement with the online programmes. This was insightful in highlighting the varying outcomes that are produced for patients who use the platform in different ways.
Again, the overall message was that any engagement with the SilverCloud programmes – no matter how often or detailed – produced positive clinical outcomes. However, it showed that there was a strong relationship between engagement and outcomes. The more engaged the user, the better their outcome and recovery.
The table*1 below breaks down the percentage improvement for the different levels of engagement.
Analysing supporter messaging
Another large dataset*2 also provided fascinating analysis of 234,735 supporter messages (supporting 54,104 SilverCloud users). The data showed that concrete, positive and supportive feedback from supporters that reference social behaviours are strongly associated with better outcomes. It showed that there is a relationship between the messages provided by the supporters - those clinicians and others who work with SilverCloud users as they explore the various programmes to tackle anxiety and depression – and increased positive clinical outcomes.
Conclusions
In totality, Dr Palacios was able to draw on analysis from 222,842 SilverCloud users, providing reliable conclusions from the three main studies he cited. The top line messages were that:
- The SilverCloud platform is highly effective at reducing symptoms of depression and anxiety, and significantly more so than similar low-intensity interventions.
- There are different ways to achieve benefit from the SilverCloud platform, but that all user engagement (no matter how frequent or lengthy) produced positive outcomes.
- Each of the insights and findings from the big data analysis helps SilverCloud’s understanding of how the platform works, for whom it works best, and under what circumstances – this is vital in continuing to develop and enhance the platform and its usage.
- These findings – and any updates to the platform which are related to these findings, can be fed back to the supporters and coaches, whilst training for these supporters is also updated in order to continue to improve the users’ experience.
We’re still in the early days of being able to use big data to help review and tailor mental health inventions. It is clear, however, that the potential insights it can provide and the ability to learn and adapt to what the data tells us is huge, with massive potential benefits from using machine learning as the three studies showcased at DTx Europe showed.
Despite the many advantages and huge potential, we must navigate the data and journey ahead with caution as each step leads onto the next – always using the data wisely and making sure that evidence guides our way forward for online CBT and all mental health interventions.
*1 https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2768347
*2 https://dl.acm.org/doi/pdf/10.1145/3313831.3376341
