You are viewing this site in staging mode. Click in this bar to return to normal site.

Data-driven refinements for better behaviour change interventions

Using frameworks to guide the development of health behaviour interventions is essential in identifying what needs to change, understanding the underlying cause of unhealthy behaviours, and considering relevant options in trying to improve those behaviours. 

However, there are few frameworks available that specifically guide the development of interventions using electronic or mobile technologies to reach and deliver health behavioural support to individuals. Recently, as part of the HPPHN Appiness event (Apps for improving the public’s health and well-being), we presented on how two frameworks – ‘Intervention Mapping’ and ‘Behavioral Intervention Technology’ – helped the design of an app to improve individuals’ active and sedentary behaviours. There were a variety of presentations and interesting discussions, among which some highlighting how the vast amounts of data generated by these technologies can provide insights to improve current practice.  

Data-driven approaches are not a novel idea, but are getting momentum in the behavioural sciences. This is because more than knowing whether or not an intervention works, we also need to understand what makes it work. Frameworks such as the ones mentioned above make important contributions towards understanding why interventions work, given that by using them, the active ingredients of interventions are being specified.

Interventions delivered electronically can greatly contribute to advance behavioural science given their fit to facilitate data-gathering on what works for whom in what circumstances. To be more exact:

  • Knowing what is delivered to who and when is easier to achieve with interventions delivered electronically, as all the “ingredients” delivered can be tracked;
  • A better understanding of contextual characteristics, such as for example the setting in which an intervention is delivered, is easier to achieve when using GPS sensors incorporated in mobile devices used to deliver the intervention;
  • Knowing individuals’ response to the “ingredients” delivered (e.g. did the individual attend to it? what was the subsequent behaviour?) is easier to achieve harnessing mobile devices’ sensing capabilities.

The large amounts of data captured with current e- & mHealth interventions can help to disentangle the factors and intervention ingredients that explain effectiveness. These data can be used to iteratively, “on the go”, refine the behavioural support delivered, its acceptability, usefulness, and ultimately its effectiveness.  However, making sense of all this data and use it to deliver more effective interventions will require artificial intelligence and machine learning approaches. 

Existing technologies and data streams already use such approaches to make individuals more aware of opportunities to purchase / consume x product, as well as to persuade into actual purchase. Can we use the same technological potential to make individuals more aware of opportunities to engage in healthy behaviour, and actually perform healthier behaviours?

Acknowledgments:

I would like to thank Prof. Corneel Vandelanotte for providing input on a draft version of this blog.

 

Dr Artur Direito, Research Associate, UCL Centre for Behaviour Change

Email: a.direito@ucl.ac.uk