Modern internet-based health interventions generate a wealth of real-time user-data that carry the promise to enable the development of personalized interventions that are highly responsive to in-the-moment patient needs. Log file analysis and Ecological Momentary Assessment (EMA) have proven to be a viable approach to tap the behavioural, emotional and social dynamics of mental health. The development of Ecological Momentary Interventions (EMI), treatments that automatically adapt on the basis of real-time user data, appears to be a logical next step. EMI, however, is still in its infancy. We have only begun to understand how to turn large and often ’messy’ datasets into clinically relevant predictive models. What is clear though, is that EMI-development requires the adoption of data mining methods that go way beyond the traditional statistical toolbox of the clinical researcher. In this 100-minute symposium, participants will learn the strengths and weaknesses of a variety of such analytical methods, through five presentations of several explorative EMA/EMI and „big data” research projects, followed by an expert comment and a plenary discussion.
Heleen Riper (VUA) introduces the theme of the symposium by providing a position statement on the need for EMI and by presenting the results of the EU ICT4Depression project, in which a team of experts from Clinical Psychology and Artificial Intelligence aimed to develop a platform for EMI research. Next, Joost Asselbergs and Jeroen Ruwaard (VUA) present results of an explorative study assessing the feasibility of smartphone-based unobtrusive EMA in predicting mood fluctuations. Third, Mark Hoogendoorn (VUA) introduces an adaptive predictive modelling framework that allows for automated clinical reasoning by combining formalized domain knowledge with individual patient data. Fourth, Johannes Smit (VUMC), reports on the use of ‘big data’ - including EMA - in mental health treatments. Fifth, Burkhardt Funk (LU) will show how eMental Health researchers can make their EMI more robust by adopting common data mining methods from the field of Artificial Intelligence. Finally, discussant David Mohr (NW-CBITS) will reflect on the presented projects, to boot up an interactive discussion with the audience.
Abstracts: 0087, 0104, 0106, 0117, 0233
Improvement of Self-Help and Blended Care Therapies for Depression by Utilizing Predictive Models Developed with Artificial Intelligence Techniques
Reinier Kop1, Mark Hoogendoorn1, Michel Klein1, Pim Cuijpers1, Heleen Riper1, Matthias Berking2, 1VU University Amsterdam, Amsterdam, The Netherlands, 2University of Erlangen-Nürnberg, Erlangen, Germany
Does Your Smartphone Know Your Mood? Preliminary Findings With Unobtrusive Ecological Momentary Assessment
Jeroen Ruwaard1, Joost Asselbergs1, 1Vrije Universiteit, Amsterdam, The Netherlands
The Promise of Big Data for Mental Health Care
Johannes H Smit1,2, Heleen Riper1,3, 1GGZ Ingeest, Amsterdam, The Netherlands, 2VU Medical Center/ department of Psychiatry, Amsterdam, The Netherlands, 3Vrije Universiteit / Department of Clinical Psychogology, Amsterdam, The Netherlands
Data Mining Applications in EMI - Challenges and Recent Applications
Burkhardt Funk1, 1Leuphana Universität Lüneburg, Lüneburg, Germany
Quantifying the Self, the Other and the Community for Depression Treatment
Heleen Riper1,5, Jeroen Ruwaard1, Jan Smit2, Pepijn Van der Ven4, Gerhard Andersson3, Pim Cuijpers1, 1VU University Amsterdam, Amsterdam, The Netherlands, 2GGZ inGeest, Amsterdam, The Netherlands, 3Linkoping University, Linkoping, Sweden, 4University of Limerick, Limerick, Ireland, 5Leuphana University, Luneburg, Germany
