Physical activity has far-reaching health benefits and contributes to the prevention of noncommunicable diseases like cardiovascular disease, cancer, and diabetes. Today's level of physical activity; however, is below the recommendations of e.g. the World Health Organization for all age groups. This amount of physical inactivity (i.e. not meeting physical activity guidelines) contributes to the rising cases of noncommunicable diseases and is responsible for over 7% of all-cause deaths along with a huge economic toll on the society. Recently, the COVID-19 crisis aggravated matters as many opportunities to be physically active were limited and sports clubs were temporarily closed. Today, effective interventions with a large reach are required to facilitate health behavior change towards more physical activity in the population. Here, even minor changes towards a more physically active lifestyle e.g. going for a daily ten-minute walk or interrupting prolonged physical inactivity can accumulate valuable health benefits over time.
There are a variety of evidence-based interventions for different settings which range from individual or group-based face-to-face interventions to digital interventions. ... mehrWhile the former is well established in today's physical activity promotion, especially for rehabilitation, the latter is especially promising to promote physical activity on a broad scale due to the availability, fast-evolving technological progress, and ease of use of digital devices in modern society. Digital interventions for health behavior change can be delivered on desktop personal computers (e.g. via DVD), over the internet (e.g. on websites), or on mobile devices (e.g. via text message or mobile application). As nearly every household worldwide has access to and experience with at least one of those devices, the potential reach and cost-efficiency of such interventions are promising. Here, the use of information and communication technologies for health, in general, is defined as electronic health while every health practice supported by mobile devices is defined as mobile health. Recently, technological advances lead to the development of smaller, more convenient, and accurate devices to continuously measure physical activity (e.g. energy expenditure, step count, and classification of physical exertion), physiological (e.g. heart rate, blood sugar, and cortisol), and report psychological (e.g. valence, energetic arousal, and calmness) parameters. This opens up new perspectives using multilevel modeling in longitudinal designs to distinguish between within- and between-person effects and allows for a higher grade of individualization of interventions. One intervention type which greatly benefits from these continuous measurements and the technological advances is just-in-time adaptive interventions. These interventions aim to deliver interventional content (e.g. motivation to be physically active) during the most promising time for the desired health behavior (i.e. physical activity) or during the most vulnerable time for unhealthy behavior (i.e. inactivity) and aim to maximize the usefulness of the intervention while minimizing participant burden. To do so, they rely on high-resolution data to depict opportune moments to deliver the intervention content. Recent progress with machine learning processes also benefits just-in-time adaptive interventions by offering sophisticated decision-making algorithms which can be guided by participants' behavior and preferences.
Previous studies on electronic and mobile interventions found heterogenic results for the effectiveness of digital health interventions for physical activity promotion. Here, evidence- and theory-based interventions which are guided by behavior change techniques (e.g. goal-setting or demonstration of behavior) were associated with higher intervention effectiveness. Furthermore, including the social context (e.g. peers, school, work, or family) in the interventions can be beneficial but it is important to distinguish between e.g. collaborative vs competitive settings based on participants' preferences. Finally, a high degree of individualization delivered by e.g. just-in-time adaptive interventions can enhance the effectiveness of mobile health interventions. However, the importance of the different interventional and contextual facets along with additional influences on the evaluation of the effectiveness remains unclear in the fast-developing field of electronic and mobile health behavior change interventions for children, adolescents, and adults.
To help close the gap between technological advances and the state of the research in electronic and mobile health interventions for physical activity promotion, this thesis aimed to 1) provide an overview of the effectiveness of electronic and mobile health interventions regarding physical activity promotion and 2) delve into important considerations and research gaps depicted by the overview (i.e. the choice of a measurement tool for physical activity and just-in-time adaptive interventions).
In our first paper, we conducted an umbrella review to summarize the evidence on the overall effectiveness of electronic and mobile health interventions along with the association of the key facets of theoretical foundation, behavior change techniques, social context, and just-in-time adaptive interventions with effectiveness. Derived from the eleven included reviews (182 original studies) we found significant benefits in favor of the intervention group (vs. control or over time) in the majority of interventions (59%). Here, the use of theoretical foundations and behavior change techniques were associated with higher effectiveness, the social context was often reported but not evaluated and just-in-time adaptive interventions were not included in any of the studies. One frequently reported shortcoming was the difficulty do compare self-reported and device-based measured results between studies. These findings suggest the potential effectiveness of digital interventions which is very likely facilitated by the key facets. Moreover, these findings helped us to determine promising but understudied facets of intervention effectiveness (i.e. just-in-time adaptive interventions) and depict frequently reported methodological issues (i.e. comparability of different measurement tools) which we could address within our thesis.
In our second paper, we explored the reliability, comparability, and stability of self-reported (i.e. questionnaire and physical activity diary) vs. device-based measured physical activity (i.e. analyzed using 10-second and 60-second epochs) in adults and children. We included two independent measurement weeks from 32 adults and 32 children in the control group of the SMARTFAMILY trial to investigate if the differences between measurement tools were systematic over time. Here, participants wore an accelerometer on the right hip during daily life and completed a daily physical activity diary for seven consecutive days. Additionally, the international physical activity questionnaire was completed by participants at the end of each week. Results indicated non-systematic differences between the measurement tools (up to four-fold). Higher associations between the measurement tools were found for moderate than for vigorous physical activity and the results differed between children and adults. These results confirm the importance of carefully considering the measurement tool to be suitable for the research question and target group and the very limited comparability between different measurement tools. Additionally, the differences within accelerometer-derived results (10-second epochs vs. 60-second epochs) point to the need for comprehensive reporting for each measurement tool to compare and replicate the results.
In our third paper, we summarized previous frameworks of just-in-time adaptive interventions and pointed out opportunities and challenges within this research field. We combined recommendations of three previous frameworks and refined that just-in-time adaptive interventions should 1) correspond to real-time needs; 2) adapt to input data; 3) be system-triggered. This can be enhanced by 4) be goal-oriented; and 5) be customized to user preferences. By doing so, just-in-time adaptive interventions can achieve a high degree of individualization which is closely fitted to each individual. The main challenge hereby remains the opportune moment identification (i.e. the exact moment when participants are either likely to engage in unhealthy behavior or when they face opportunities to perform healthy behaviors) to timely deliver intervention content. This can be explored using ambulatory assessments and assessing the context of the behavior. The decision-making process can be enhanced by machine learning algorithms. These results guided the reporting and design of the examinations included in our fourth and fifth papers.
In our fourth paper, we evaluated the importance of engaging with a just-in-time adaptive intervention triggered after a period of physical inactivity. For this secondary data analysis, 47 adults and 33 children were included in the analysis who wore an accelerometer on the right hip and used our SMARTFAMILY2.0 application during the three-week intervention period of the SMARTFAMILY2.0 trial. Here, we analyzed 907 just-in-time adaptive intervention triggers and compared step and metabolic equivalent count in the hour after occasions when participants answered the trigger (i.e. responded to the question regarding their previous physical inactivity) within 60 minutes ("engaged" condition) with the hour after occasions when they did not answer the trigger within 60 minutes ("not engaged" condition) in the mobile application. Results indicated significantly higher metabolic equivalent and step count for the "engaged" condition within-persons. This shows that if a person engaged with a trigger within 60 minutes, he or she showed significantly higher physical activity in the following hour compared to when the same person did not engage with the trigger. This expands previous research about participants' engagement with the intervention and the importance of an opportune moment identification to enhance this engagement.
In our fifth paper, we explored the association of sleep quality and core affect with physical activity during a mobile health intervention period. Based on the same intervention period reported in the fourth paper, but with different inclusion criteria for the data (e.g. minimum wear time of the accelerometer for 8 hours per day instead of 80% of the hour of interest), daily accumulated self-rated mental state was compared to step count and minutes of moderate-to-vigorous physical activity for 49 adults and 40 children in a secondary data analysis. Overall, 996 measurement days of the participants were included in this analysis. Our results showed that higher reported valence and energetic arousal values were associated with more physical activity, while higher reported calmness values were associated with less physical activity within-persons on the same day. No distinct association was found between sleep quality and physical activity. Our results confirm previous ambulatory assessment studies and we suggest that within-person associations of core affect should be considered when designing physical activity interventions for both children and adults. Additionally, core affect might be a promising consideration for opportune moment identifications in just-in-time adaptive interventions to evaluate the feasibility and causality of targeting changes in e.g. valence to improve subsequent and daily physical activity of participants using micro-randomized trials.
Based on the current state of knowledge, our results above address important research gaps depicted by our overview in the field of digital interventions for physical activity promotion. One example is the understudied area of just-in-time adaptive interventions for which we provided a framework, evaluated the effect of engaging with such interventions on subsequent physical activity, and explored core affect and sleep quality as facilitators of physical activity behavior. With these findings in mind, we discussed important considerations to progress future mobile health studies for physical activity promotion in general, and just-in-time adaptive interventions in particular at the end of this work. Finally, we aimed to transfer this knowledge into a proposal for designing a just-in-time adaptive intervention in the special group of participants at risk for or with knee osteoporosis who could specifically benefit from this highly individualized approach.
1