Computer-tailored Digital Health: the past, present & future
By Artur Direito, on 15 November 2017
By Dr. Eline Smit, Assistant Professor at the Amsterdam School of Communication Research/ASCoR, University of Amsterdam
Ever since I started my PhD project on the effectiveness and feasibility of a web-based computer-tailored intervention aimed at smoking cessation, digital health interventions have been one of my main research foci. Driven by my curiosity and eagerness to move this field of research forward, at the most recent conference of the European Health Psychology Conference (EHPS) I organised a symposium on ‘Innovative ideas in online computer-tailoring’. In this blog post, I would like to elaborate on my introductory talk, in which I talked about the past of online computer-tailoring, its present and future.
For those of you that are unfamiliar with computer-tailoring as a behaviour change technique, computer-tailored digital health interventions can be best described as interventions that are adjusted in terms of their content based on the specific characteristics of the individual participant, using a computerized process. This is expected and has been proven to increase personal relevance of the intervention, information processing, as well as use of and engagement in the intervention. At this very moment, we can even conclude with a fair amount of confidence that computer-tailored digital health interventions can be effective in changing a variety of health behaviours.
This proven effectiveness has led to several ongoing trends within the computer-tailored digital health arena and I would like to highlight two of them:
- The extension of research on effectiveness, applying the idea of computer-tailoring to different target groups (e.g. intermediaries, such as health professionals – elaborated upon in the presentation by Dennis de Ruijter), and to different types of behaviour (e.g. intermediate behaviours, like the decision to use smoking cessation support tools).
- The investigation of cost-effectiveness, which is easier said than done due to several challenges that come with the economic evaluation of digital health interventions. Yet, while doing the best we can with our current knowledge and skills, research findings so far appear to be rather promising.
At the same time, we need to stay critical and realize that the effect sizes of computer-tailored digital health interventions remain relatively small, limiting their possible impact on public health. Therefore, the other three presentations in the symposium concerned several innovative ideas on how to increase the effectiveness of computer-tailored digital health interventions, thereby moving the field forward:
- Felix Naughton presented on the idea of context-tailoring, enabling real-time behavioural support to be triggered and tailored to changes in a user’s situational context using smartphone sensors;
- Maria Altendorf presented on the idea of message frame tailoring, adjusting the message frame in which the information is presented based on people’s need for an autonomous or more directive communication style;
- Hao Nguyen presented on the idea of mode tailoring, adjusting the mode of information delivery based on an individual’s preferences for text, text with illustrations, text with video, or a combination.
Whereas the results on mode tailoring show that this could be a promising strategy to increase the effectiveness of digital health interventions, results on the effectiveness context-tailoring and message frame tailoring are yet to be waited for.
Food for thought
We ended the symposium with some thought-provoking questions, nicely formulated by Ciska Hoving. Given the enthusiasm of the people attending as well as their willingness to ask questions themselves, however, not all of them were in fact addressed. Therefore, this blog post provides me with the perfect opportunity to pose these questions again – your responses are very much welcomed!
- What is the worth of tailoring type (e.g. context, mode, message frame) in the grand scheme of computer tailoring effectiveness?
- How can we balance intervention effect and ‘typical’ use of online sources?
- What is the practical future of computer tailoring?
Eline is an assistant professor at the Amsterdam School of Communication Research/ASCoR, University of Amsterdam, and part of the Amsterdam Center for Health Communication (for her profile, click here). Her main research interests concern tailored health communication delivered through the Internet. Specifically, her research focuses on the integration of innovative strategies such as eHealth interventions into the healthcare setting and the exploration of novel tailoring strategies, such as message frame tailoring. She has an extensive track record of peer-reviewed articles and has obtained several grants for research projects in this field. Last August, she received the prestigious Early Career Award at the annual conference of the EHPS. You can follow Eline on Twitter (@smit_eline).
2 Responses to “Computer-tailored Digital Health: the past, present & future”
Eline wrote on 17 November 2017:
Thank you very much for your enthusiastic reply!
I totally agree with you that all the work that none of these novel methods of tailoring that we’re investigating might in fact be the holy grail.
I very much like your ideas on the integration of different objectives sources of data, thereby also greatly reducing the burden we’re usually placing on our respondents with large questionnaires. I’d be happy to talk about this further, so will definitely get in contact with you soon!
Very interesting piece! I agree with your assessment that computer-tailored interventions are effective, but not effective enough. I also agree that all the suggestions you make (context, mode, message-frame) have the potential to improve the effectiveness of these interventions. I have actually done a fair bit of work on using different delivery modes (video, text, combined). However, I also think that neither of these solutions will provide the big breakthrough that will make computer-tailored interventions a lot more effective. More like incremental progress…nothing bad with that…but I’m pretty sure that the quest to make these intervention more effective will still be on after incorporating these suggestions.
In my mind we need to make our interventions more personal, less artificial, better integrated into peoples daily lives, more adaptable and responsive,…all this while not overburdening participants with questions to ask (more variables to tailor too), and without the need to develop even more extensive message libraries (there needs to be a limit somewhere). We need a whole new way of thinking about how to provide personalised behaviour change content.
I see a great potential of integrating a range of objective datasources, delivering minute-by-minute data ( from bio-feedback from (activity) trackers, weather information, GPS/GIS, calendar/diary/travel data,…) and to make sense of this continuous stream of data, we will need some amazing computing power…AI/deep learning come to mind… all the while not forgetting to integrate essential knowledge about the behaviour and addressing all the relevant BCTs.
I don’t even know how you would begin to build such intervention, very different from the classic computer-tailored intervention… but someone will eventually! If anybody is out there thinking along the same lines, I’d be more than happy to talk to you just send me a message (email@example.com).