Towards more personalized digital health interventions: a clustering method of action and coping plans to promote physical activity

Abstract

Background Despite effectiveness of action and coping planning in digital health interventions to promote physical activity (PA), attrition rates remain high. Indeed, support to make plans is often abstract and similar for each individual. Nevertheless, people are different, and context varies. Tailored support at the content level, involving suggestions of specific plans that are personalized to the individual, may reduce attrition and improve outcomes in digital health interventions. The aim of this study was to investigate whether user information relates toward specific action and coping plans using a clustering method. In doing so, we demonstrate how knowledge can be acquired in order to develop a knowledge-base, which might provide personalized suggestions in a later phase. Methods To establish proof-of-concept for this approach, data of 65 healthy adults, including 222 action plans and 204 coping plans, were used and were collected as part of the digital health intervention MyPlan 2.0 to promote PA. As a first step, clusters of action plans, clusters of coping plans and clusters of combinations of action plans and barriers of coping plans were identified using hierarchical clustering. As a second step, relations with user information (i.e. gender, motivational stage, ...) were examined using anova’s and chi2–tests. Results First, three clusters of action plans, eight clusters of coping plans and eight clusters of the combination of action and coping plans were identified. Second, relating these clusters to user information was possible for action plans: 1) Users with a higher BMI related more to outdoor leisure activities (F = 13.40, P < .001), 2) Women, users that didn’t perform PA regularly yet, or users with a job related more to household activities (X2 = 16.92, P < .001; X2 = 20.34, P < .001; X2 = 10.79, P = .004; respectively), 3) Younger users related more to active transport and different sports activities (F = 14.40, P < .001). However, relating clusters to user information proved difficult for the coping plans and combination of action and coping plans. Conclusions The approach used in this study might be a feasible approach to acquire input for a knowledge-base, however more data (i.e. contextual and dynamic user information) from possible end users should be acquired in future research. This might result in a first type of context-aware personalized suggestions on the content level. Trial registration The digital health intervention MyPlan 2.0 was preregistered as a clinical trial (ID:NCT03274271). Release date: 6-September-2017.

Towards more personalized digital health interventions: impact of emotions and physical complaints on goals, action and coping plans to promote physical activity

“…However in order to develop digital interventions with more dynamic personalized suggestions, these associations should be embedded in a sort of computerized system to actually recommend suggestions to participants (Çelik Ertuğrul & Elçi 2020). Previous work from our research group already stated that the use of a knowledge-based system would be appropriate (Schroé et al 2022a). A knowledge-based system is a computer program that contains a knowledge base which represents knowledge in a particular domain (Pratiwi et al 2018) and reasons with that information via intelligent algorithms in order to deliver personalized suggestions.…”

“…Previous work already investigated how personal user information (e.g. gender, age, motivational stage, etc) related towards the content of action and coping plans using a clustering method (Schroé et al 2022a). Although this approach was promising, this data was relatively stable over time and would only result in personalized suggestions at the level of the user pro le.…”

“…We judged that there was no logical categorization for the Self-Enactable Techniques. Second, solutions are already likely to be associated with the barriers in itself (Schroé et al 2022a), so we decided after careful internal consideration that investigating the association between emotions/physical complaints and solutions was less important. The quality of the action plans and barriers was also coded in terms of instrumentality (e.g.…”

“…Notwithstanding, this resulted in a low quality of plans and participants also experienced di culties in formulating them (Degroote et al 2020; Poppe et al 2018). As such, there is a need to tailor this support in a more personalized and dynamic way (Schroé et al 2022a).…”

The relationship of affective and bodily states with goals and plans to increase physical activity: An 8‐day study in students

“…To our knowledge only one study looked at characteristics of individuals that may predict the specific content of the PA plans made. Schroé et al (2022) found that female individuals, those with lower general levels of PA and those that were employed were more likely to create plans for household activities. Younger users were more likely to plan active transport and sport activities.…”