EMASENS
Developing a just-in-time adaptive intervention to prevent smoking lapses
The aim of the EMASENS project is to develop a smartphone-based stop smoking tool which provides personalised support to smokers in real-time, when they most need it (i.e., a ‘just-in-time adaptive intervention’). To provide such personalised support, we first need to understand when to send support and what type of support to send.
In the first stage of the EMASENS project, we aimed to gain a better understanding of when and why people are at risk of lapsing (i.e., smoking cigarettes after the quit date). To build a lapse prediction algorithm, we triangulated findings from two studies using slightly different approaches.
In the first study, we used observational data from a popular smoking cessation app to develop supervised machine learning algorithms to distinguish lapse from non-lapse reports. We used data from app users with ≥20 unprompted data entries, which included information about craving severity, mood, activity, social context, and lapse incidence. A series of group-level supervised machine learning algorithms (e.g., Random Forest, XGBoost) were trained and tested. Their ability to classify lapses for out-of-sample i) observations and ii) individuals were evaluated. Next, a series of individual-level and hybrid algorithms were trained and tested. We found that using unprompted app data appeared feasible for constructing a high-performing group-level lapse classification algorithm but its performance was variable when applied to unseen individuals. Algorithms trained on each individual’s dataset, in addition to hybrid algorithms trained on the group plus a proportion of each individual’s data, had improved performance but could only be constructed for a minority of participants. We therefore deemed it important to repeat the analyses using data from a prompted study design (using both signal- and event-contingent Ecological Momentary Assessments), with results triangulated with those from the present study to arrive at a better understanding of when and why lapses occur.
In the second study, we used self-reported data from brief, hourly surveys sent directly to people’s smartphones in addition to passively collected data from smartwatches to train and test group-level, individual-level, and hybrid (i.e., a combination of group- and individual-level) lapse prediction algorithms. We found that individual-level and hybrid algorithms performed better than the group-level algorithms, particularly when including the passively collected sensor data. However, multiple success criteria (e.g., acceptability, scalability, technical feasibility) need to be carefully balanced against algorithm performance in the JITAI development and implementation process.
The key output from the first stage of the EMASENS project was the selection of a best-performing algorithm to take forwards to underpin a smoking cessation JITAI.
In the second stage of the EMASENS project, we will develop the intervention messages using theory and behaviour change tools (e.g., the Behaviour Change Wheel) in combination with extensive user testing. A bank of intervention messages will be developed, tailored to each identified lapse risk factor. Next, the effectiveness of the JITAI will be evaluated. The second stage of the EMASENS project forms part of Corinna Leppin’s PhD project.