Dr Olga Perski is a Senior Research Specialist at Karolinska Institutet, Sweden, where she leads the Precision Health Psychology Research Group. Her work focuses on the development, optimisation and evaluation of digital interventions to help people quit addictive behaviours, with a focus on real-time assessment of within-person processes and just-in-time adaptive interventions. Since completing her PhD in Health Psychology at University College London in 2018, she has worked in the UCL Tobacco and Alcohol Research Group and held a Marie Skłodowska-Curie Postdoctoral Fellowship at the University of California, San Diego and Tampere University. She currently leads a five-year ERC-funded Starting Grant that uses high-resolution smartphone and smartwatch data to better understand and prevent relapse across multiple addictive behaviours using methods from the control systems engineering toolbox.

Abstract

Focusing on the person, not the substance: Combining theory- and data-driven methods to build transdiagnostic relapse prevention just-in-time adaptive interventions

Just-in-time adaptive interventions (JITAIs) for relapse prevention aim to deliver the right type and amount of support to each person, when and where they most need it. To achieve this aim, JITAIs need an underlying model of when and why an individual is vulnerable to lapse and relapse. Addiction theory alone is not specified at a level that can tell us when and where to provide support and purely data-driven approaches often have poor interpretability and typically do not generalise beyond the specific sample and addictive behaviour they were trained on.

In this talk, Olga will outline an approach which combines theory- and data-driven methods from the control systems engineering toolbox. Formal and computational modelling helps specify dynamic mechanisms and JITAI decision points. These models can then be tuned using system identification experiments and intensive longitudinal data and used for prediction and “automated control”. She will also make the case for moving beyond single-behaviour JITAIs. As addictive behaviours share dynamic mechanisms, transdiagnostic JITAIs that adapt to the person rather than the specific substance or behaviour can potentially enhance scalability and clinical utility.