Comparing Familiar with Inspiring Recommendations to Assist People in Moving More Ine Coppens1,* , Luc Martens1 and Toon De Pessemier1 1 WAVES - imec - Ghent University, iGent - Technologiepark-Zwijnaarde 126, Ghent, Belgium Abstract Sufficient physical activity is crucial for people’s health and well-being. However, not enough people attain the weekly minimum of 150 minutes. Since current mobile health systems are not optimal to motivate and assist people to move more, this study investigates the effect of personalized suggestions generated by two types of recommender system algorithms: content-based (which provide more familiar recommendations, relevant to existing interests) and user-based collaborative filtering (which deliver more diverse recommendations, allowing inspiration for new interests). By conducting a longitudinal between-subject user study over eight weeks, we will investigate how the two algorithms separately affect motivation and behavior change. We developed two versions of an Android smartphone application to deliver the recommendations, with the only difference being the implemented recommender algorithm. In all other aspects, the apps are identical: Both systems use the same datasets of physical activities and tips to break sedentary behavior, apply the user profile and contextual filter, and integrate the combination of star rating and momentary motivation feedback to provide personalization on preferences and well-being. We will analyze the differences in people’s star rating feedback, motivation to move, physical activity, and sedentary behavior. The main hypothesis is that inspiring recommendations from the collaborative algorithm will motivate people more for more physical activity and less sedentary behavior. The results of this study will provide insights for future mobile health recommenders in what type of algorithm and recommendations are most effective in the domain of increasing physical activity and motivating people to move more. Keywords health recommender system, physical activity, motivation, behavior change, mobile health, assistive healthcare, sedentary behavior 1. Introduction to break SB [6], more general healthy habits [7], or reminders and tips [8]. Despite their great potential to Insufficient physical activity (PA) is one of the modifiable motivate people, the interventions are often underused underlying causes of chronic diseases, which cause most [9]. Furthermore, other research suggests that they deaths worldwide [1]. The World Health Organization currently have limited effects on PA and SB, even when (WHO) defines evidence-based guidelines for increasing implementing behavior change techniques, such as goal PA and reducing sedentary behavior (SB) [2]. However, setting and self-monitoring [10]. This implies the need in 2016, 27.5% of the adult population did not meet their for new technologies and more interactive interventions recommended minimum of 150 minutes of moderate aer- [10]. obic PA per week [1]. Since sufficient PA is essential for To increase user engagement and behavior change to- people’s health and mental well-being, PA promotion is wards more PA, mHealth systems can implement Recom- now more crucial than ever [3]. mender System (RS) algorithms to deliver personalized Electronic health (eHealth) and mobile health and relevant interventions to the user [9, 11]. RSs gener- (mHealth) interventions use technologies to promote ate personalized suggestions based on user preferences healthy behavior [4]. As such, they can also be used to help them with making decisions [12]. They can also to assist people in moving more by promoting PA be applied in the health domain as Health Recommender and prevent long periods of SB. In previous eHealth Systems (HRSs) to propose healthier suggestions, tailored and mHealth studies to increase PA, the content of to the user [13]. Previous work has applied RS techniques their interventions ranged from activities [5], ideas to provide personalized well-being recommendations for Ine Coppens, Luc Martens and Toon De Pessemier. 2023. Comparing food and PA [14], for personalized training sessions for Familiar with Inspiring Recommendations to Assist People in Moving marathon running [15], and for health activities [11]. Al- More. In Joint Proceedings of the ACM IUI 2023 Workshops, March though providing the most relevant health suggestion to 2023, Sydney, Australia, 8 pages. the user would optimize mHealth interventions, appli- * Corresponding author. cation of HRSs for behavior change is still in its infancy $ Ine.Coppens@UGent.be (I. Coppens); Luc1.Martens@UGent.be (L. Martens); Toon.DePessemier@UGent.be (T. De Pessemier) [9, 13].  0000-0002-3051-506X (I. Coppens); 0000-0001-9948-9157 To generate useful recommendations, the RS has to (L. Martens); 0000-0002-3920-7346 (T. De Pessemier) predict what the relevant items are for the user, for which © 2023 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). several techniques exist [9, 12]. The content-based tech- CEUR Workshop Proceedings http://ceur-ws.org ISSN 1613-0073 CEUR Workshop Proceedings (CEUR-WS.org) 1 Joint Proceedings of the ACM IUI Workshops 2023, March 2023, Sydney, Australia nique suggests items similar to items liked by the user in behavior change by having more PA and less SB, as recom- the past, based on attributes that describe the items [12]. mended by the WHO [2]. Because the two RS techniques Alternatively, collaborative filtering uses other users’ rat- will provide different recommendations, we expect a dif- ings and assumes that people with the same interests ferent effect on user motivation and behavior. To the will like the same items [9]. The item-based collaborative best of our knowledge, this effect of different types of RS filtering technique recommends similar items based on algorithms has not been investigated. As such, we exam- the ratings of other users, while the user-based collabora- ine which RS algorithm will perform best in motivating tive filtering technique focuses on recommending items users for more PA and less SB, responding to the demand that similar users with similar preferences liked in the of enhancing health interventions with the best personal- past [12]. While the item-based method provides more ization approach [9]. By developing two versions of the accurate recommendations as the user’s preferences are same Android app, we will conduct a between-subject modeled using similar items, the user-based approach user study with the following research question: can recommend more diverse and unexpected items [12]. Which recommender algorithm has the best effect on Providing different approaches to predict what a user people’s star rating feedback, motivation to move, physical might like, these RS techniques result in a different selec- activity, and sedentary behavior? tion of recommended items [9]. While content-based RSs succeed in recommending highly relevant items, they of- ten suffer from overspecialization as they suggest items 2. Methods very similar to items the user already knows because the We developed two HRSs that recommend personalized attributes are already defined in the user profile [16]. As PAs and tips for breaking SB to assist users in their daily such, they fail at recommending more unexpected, sur- life in moving more. For these PA and tip items, our own prising, and novel items that could still be relevant to the two datasets were created. The PA dataset was assembled user [12, 17]. Previous work has addressed this overspe- using 354 PAs from the Compendium of Physical Activi- cialization problem on the grounds that it leads to lower ties [21]. The tip dataset contains ideas from the Belgian user satisfaction [18, 17, 19, 16]. Collaborative systems website for health (www.gezondleven.be/), resulting in solve this problem because they can recommend items 81 items. The generated recommendations are delivered with a very different content when it is liked by similar to the user in an Android app called MoveMoreApp, as users [12, 16]. To summarize, there are content-based shown in Figure 1(a), with its interface similar to our algorithms that provide familiar recommendations which app from a previous study. This app shows three PA are highly relevant to existing interests, and collaborative and three tip recommendations. When a user executes filtering that can deliver more diverse and unexpected an item, manual feedback on the recommended items is suggestions which allow new interests to be explored collected as a rating on five stars, as illustrated in Figure [12, 19, 17]. Hybrid RS algorithms combine the advan- 1(b) with the question “how do you rate the generated rec- tages of the content-based and collaborative approach, ommendation?”. Additionally, our system collects users’ providing a balance between relevant and diverse recom- momentary motivation to move with a slider measured mendations [12, 18]. on a 5-point Likert scale (from “not motivated” to “ex- In this research, however, we do not want to balance tremely motivated”), as depicted in Figure 1(c). the characteristics of the algorithms by merging them in a hybrid RS. Rather, we want to study the algorithms and the impact of their advantages and disadvantages 2.1. The algorithms separately in the domain of PAs. For example, previous The PA and tip items are recommended to the users with research has shown that repetition of the same health two types of RS algorithms, as illustrated in Figure 2. behavior makes the behavior easier [20], suggesting that The initial filter based on the user’s profile (available overspecialization may not be a problem in the domain material and maximum impact level) and the contex- of PA. Similarly, we chose to implement the user-based tual filter based on the current weather (obtained using version of collaborative filtering because it can recom- https://openweathermap.org/) and remaining daylight mend more diverse items than the item-based version are applied on the PA and tip datasets in both groups to [12], and because we want to emphasize the effect of remove unsuitable items. more diverse recommendations on people’s behavior. As In the next step, the RSs generate personalized rec- such, we investigate the content-based and user-based ommendations based on the users’ consumption his- collaborative RS algorithm separately as two extremes tory. This history contains the PAs and tips the user (very relevant versus very diverse) to gain understanding engaged in, together with the provided star rating feed- in how they each affect motivation and behavior change. back, momentary motivation, and the user’s mood. The In this study, concrete PAs and tips to break SB are rec- star rating feedback and momentary motivation are both ommended with the goal to motivate people for healthy 2 Joint Proceedings of the ACM IUI Workshops 2023, March 2023, Sydney, Australia (a) (b) (c) Figure 1: Three recommended activities and three recommended tips are shown in the user interface of MoveMoreApp (a). After selecting and having engaged in an activity or tip, users can specify more details about it, such as the location, buddy, star rating, and for PAs also feedback on intensity and duration (b). After submitting the item, the app also asks the momentary motivation for the activity or tip, together with the user’s current mood (c). measured on a scale of five and are aggregated with sumption history also contains the situation history at equal weights in the formula: 𝑎𝑔𝑔𝑟𝑒𝑔𝑎𝑡𝑒𝑑_𝑓 𝑒𝑒𝑑𝑏𝑎𝑐𝑘 = the corresponding time. To re-rank the items, a value (𝑟𝑎𝑡𝑖𝑛𝑔 + 𝑚𝑜𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛)/2. In this way, our two RSs op- between 0 and 1 that represents how close in time the timize their recommendations on both the rating and item’s situation is to the situation’s occurrences in the motivation. The mood is asked at the beginning of every history is added to the preference estimation score. As a day and after every submit with several emoji, as shown result, items that match better with the estimated current in Figure 1(c). As such, the user’s current mood is used to situation appear higher in the list of recommendations. filter the consumption history on previous consumptions Next, the recommended PA items go through the adap- with a similar mood, based on the mood micro-profile of tive step. Combined with the user’s current PA level and [22]. feedback on intensity and duration provided in the app, The content-based RS algorithm only needs the user’s as shown in Figure 1(b), the system provides a gradual in- own consumption history. Calculating the similarity to crease in PA intensity and duration, following guidelines items consumed in the past relies on attributes that de- of the WHO [2] and the European Society of Cardiology scribe the items [12]. As such, our PA and tip dataset were [3]. In the final step, the recommended PAs and tips are extended with corresponding attributes to describe each shown to the user. item, such as aerobic, flexibility, or balance. The content- Right at the beginning, when the users did not submit based algorithm uses these to represent the user’s pref- and rate any PAs or tips yet, there is no consumption his- erences and match these with all the filtered PA and tip tory present to derive the user preferences from and base items using the cosine similarity [12]. In the other group, the recommendations on, resulting in the new user cold the collaborative filtering searches for similar users who start problem [12]. To provide an initial recommendation provided similar feedback to the same items using the with the available information, the two algorithms apply cosine similarity, and calculates a preference estimation the user profile filter and the contextual filter, and then score for all the filtered PAs and tips [12]. randomly select PAs and tips from this filtered set. As At this point, both RS algorithms generated a list of more PAs and tips are chosen, the consumption history recommended items with corresponding preference es- will grow over time, resulting in better, more personal- timation scores. The contextual post-filter re-ranks the ized recommendations. It is however possible that users items based on the current estimated situation [12]. In do not like any of the (initial) recommendations and do our study, this situation can be: free time, during work, not select anything. In that case, the app allows users to household task, or active transport, and is assigned to select their own chosen PAs from the PA dataset with every item in the two datasets. In this way, the con- a search functionality when clicking on the “enter own 3 Joint Proceedings of the ACM IUI Workshops 2023, March 2023, Sydney, Australia Content- Content- Adaptive Recommend PA Context Context Based Filter Based pre-filter post-filter Group RS Recommend Tip Users’ Own consumptions consumption history Similar users’ consumptions Collabora ve User-Based Adaptive Recommend PA Filtering Context Context Filter Collaborative pre-filter post-filter Group RS Recommend Tip Figure 2: A schematic representation of how the algorithms in both versions work shows that the only difference between the two groups is the applied RS algorithm. activity” button, as shown in Figure 1(a). These PA con- the generated recommendations and they are stimulated sumptions are used in the RS algorithm for subsequent to only submit items when actually having engaged in recommendations. The new item cold start problem, on the them, rather than only rating them for more money. other hand, occurs when no ratings for the unexplored We designed the processing of data collected by our items are available yet [12], which is not a problem for app together with our ethical committee and data pro- the content-based RS since this algorithm only uses at- tection officers to be compliant with the General Data tributes to recommend other items. The collaborative Protection Regulation (GDPR) and our study received RS however, does depend on item ratings from other ethical approval. users [12]. We address this new item cold start problem by integrating an initial user-item consumption dataset 2.3. User study design from a previous study, in which items already received ratings from other users and to which no new items are A longitudinal user study will be conducted following added. Moreover, since our item datasets are relatively a between-subject study design in which each user is small compared to the amount of users (354 PAs and 81 assigned to either the content-based or collaborative fil- tips) [12], and since we expect that users will engage in tering method. The advantage of between-subject user daily PA (which is any movement of the body, as defined studies is the possibility to investigate the long-term ef- by the WHO [2]), we estimate a sufficient amount of fect of one system separately without having to switch consumptions after one week to alleviate the cold start between systems, but it also requires more users and problems. more interactions [12]. As illustrated in Figure 2, the only difference between the two groups is the type of RS algorithm. The other steps (user profile filter, contextual 2.2. Participants pre-filter, contextual post-filter, and adaptive algorithm The target group of our study are adults who currently do for PAs) are exactly the same. not achieve the 150-minute weekly minimum of moderate Participants are asked to install the Android applica- PA. An initial screening with questions about age, weekly tion on their own smartphone. Immediately after installa- amount of PA [23], and a PA screening [24] in the app tion, the app randomizes the participants in the content- will decide whether or not the participant is eligible to based or collaborative filtering group. Then, participants join the study. Aimed at recruiting 50 participants, we are asked to answer the pre-test questionnaire, followed promote our study via the Sona research participation by an eight-week study. During these eight weeks, they system of Ghent University and several Facebook groups can use the app in their daily life to look at the recom- for paid studies. The study will run from March until mendations and choose an item to execute. When an June, 2023. item is selected, as shown in Figure 1(a), this is saved in Participants will receive an incentive of 30 EUR when the app even when the app is closed during the execu- they used the app for eight weeks and answered all the tion of the activity. When the activity or tip is executed, questionnaires. They are not rewarded for having more the user goes back to the app to submit and rate it, as PA or for the amount of PAs or tips they submit, because depicted in Figure 1(b), in which the eventual duration of they can also use the app with “not now” and “enter own the executed PA is also asked. As such, participants are activity” submits. As such, they are free to choose from requested to only submit PAs and tips after engaging in 4 Joint Proceedings of the ACM IUI Workshops 2023, March 2023, Sydney, Australia them to provide proper feedback on the eventual rating, European Health Interview Survey - Physical Activity motivation, and duration. After eight weeks, the app Questionnaire (EHIS-PAQ) [23], and SB, surveyed with shows a final post-test questionnaire. the Sedentary Behavior Questionnaire (SBQ) [29] because Since the goal of our study is to investigate the differ- they both allow participants to reflect on their average ences of receiving personalized recommendations from weekly PA and SB behavior, and they both distinguish either the content-based or the collaborative RS algo- between different situations, such as PA or SB at work rithm, the study duration is dependent on the time it or as transport. Repeated Measures ANOVA tests will takes for the RSs to succeed in generating personalized be conducted to investigate the evolution in motivation recommendations. By providing solutions for the cold regulation style and behavior change between the pre- start problems as discussed earlier, we expect that the RSs and post-test measurements and between the two groups will be able to provide personalization after one week. In [30]. total, we decided on a study duration of eight weeks, rea- A manipulation check will validate whether the ma- soning that longer durations would result in more user nipulation succeeded. The manipulation in our study dropout [9]. We expect that users will have submitted is generating either familiar recommendations with the sufficient consumptions, and that sufficient PAs and tips content-based RS, or diverse recommendations with the will have been recommended in eight weeks to answer collaborative RS. The user’s experience of these recom- our research question. mendations can be measured with the questionnaires of [31]. In these questionnaires, different RS properties are 2.4. Measures and analyses surveyed, such as perceived recommendation accuracy and quality (e.g., “The recommended items fitted my pref- When the study is finished, statistical analyses will be erence”), and additional properties that measure beyond conducted using IBM SPSS Statistics Version 28 to answer accuracy, such as perceived recommendation diversity our research question. The research question is divided and variety (e.g., “The list of recommendations was var- into four main dependent variables: star rating feedback, ied”) [31, 12]. To keep the app user friendly, the app will motivation to move, amount of PA, and SB. These vari- not ask these questionnaires every time the user receives ables are all measured using the Android app at different a recommendation. Instead, the app will randomly show points in time. Depending on the timing of measurement these questionnaires in 20% of the time after the user of the dependent variable, different types of statistical chose and submitted a PA or tip recommendation. As a tests will be conducted on the longitudinal dataset and result, these data will also be longitudinal with repeated the pre-post dataset. measures over eight weeks, and Generalized Estimating Firstly, measurements per individual are repeated over Equations [25] will be conducted for the analysis of the the eight-week study resulting in a longitudinal dataset. manipulation check. The repeated measurements include: star rating feedback As the success and usefulness of an RS algorithm is on a recommended item, momentary motivation to move, based on how well it can predict the user’s preferences and the daily executed PAs and tips. Because of this [12], the stability of the preferences determines which longitudinal data, in which the data can be unbalanced algorithm will provide the best recommendations [17]. (e.g., not every user engages in the same amount of PAs), In some domains, such as movies, user preferences are analyses will be conducted with Generalized Estimating mostly stable over time, thus eliminating the need for Equations [25] to investigate differences between the diverse recommendations [17]. On the other hand, some groups. people seek variety in their behavior, indicating the need Secondly, motivation and behavior change are also for novelty and diversity in the recommendations [12]. measured in both the pre- and post-test questionnaires to In this case, RSs should take into account the differences investigate their evolution after the eight-week study. To in user preferences, which can be depended on their measure motivation, we chose to utilize the regulation personality [12] or change over time [32]. For this reason, types of motivation as defined by the self-determination we also survey the user’s preference for variety in the theory (SDT), a theory of motivation that distinguishes pre-test questionnaire with our own questions, rated on between autonomous and controlled motivation [26]. a 5-point Likert scale from “Disagree strongly” to “Agree Based on the SDT, the motivation for PA (RM4-FM) ques- strongly”: “I like variety in my daily physical activities” tionnaire [27] and the Behavioral Regulations for Exer- and “I prefer routine in my daily physical activities”. This cise Questionnaire (BREQ) [28] measure the motivation independent variable will serve as a control variable in types for PA and exercise, respectively. By using separate the aforementioned analyses. questionnaires, we differentiate between PA, which the To evaluate the overall performance of all the steps WHO defines as any movement of the body [2], and exer- of the algorithms, the “not now” button allows users to cise, which is a subset of PA. To measure behavior change, provide a reason why now is not a good time for PA. We we chose to analyze changes in PA, surveyed with the provided our own feedback sentences to check whether 5 Joint Proceedings of the ACM IUI Workshops 2023, March 2023, Sydney, Australia or not the recommendations fit with the weather (e.g., vation, we hypothesize that the increase of PA and the “It is raining too much”) or with the current mood (e.g., decrease of SB will be stronger in the collaborative fil- “I do not feel good”), whether or not they are adapted to tering group because autonomous motivation results in the user’s PA level (e.g., “The recommendations are too more effective healthy behavior change [26]. intense”), and whether or not the situation is suited for Lastly, we hypothesize that the collaborative RS will the recommendation (e.g., “I’m still at work/school”). perform better (e.g., higher star ratings, momentary mo- tivation, and amount of PA) when combined with a user who needs more variety in their behavior because it gen- 3. Expected results erates more diverse recommendations [12] and allows exploration of new items and interests [17]. Similarly, We will first check whether our manipulation succeeded we hypothesize that the content-based RS will perform by analyzing the users’ experience with the generated better when combined with a user who prefers routine be- recommendations. We expect that participants in the cause it generates recommendations similar to items the content-based group will assign larger scores for per- user already engaged in and already knows [12, 17, 16]. ceived recommendation accuracy and quality [31] be- Moreover, repeating the same behaviors can make them cause the content-based algorithm will generate recom- easier [20], mitigating the overspecialization problem mendations that fit better with the user preferences [12]. of the content-based RS. Since this research examines Furthermore, we check whether the collaborative algo- whether RSs should focus on existing interests or on rithm provided more diverse recommendations, as we discovering new interests in the domain of PAs in an expect larger scores for perceived recommendation di- eight-week period, we will not investigate whether or versity and variety [12, 31]. not these interests persist as habits, as previous research As content-based RSs generate recommendations that has indicated that habit formation may take up to 254 are similar to previously consumed items, and thus, fit days [20]. better with user preferences [12], we hypothesize that the assigned star rating feedback will be higher in the content-based group. However, content-based RSs do not 4. Conclusions and future work provide an exploration of new items and expansion of their knowledge [17], and they ignore items with little This research investigates whether content-based or col- similarities [18]. Moreover, we expect that integrating laborative filtering recommendations have a better effect more variety and unexpected items in the recommenda- on people’s motivation and behavior change for PA when tions with collaborative filtering will enhance their en- implemented in an HRS that assists people in moving joyment [9], inspire them with new interests, and expand more. The effectiveness of the HRSs will be evaluated their horizon [12, 17]. We hypothesize that increasing with a between-subject eight-week user study and an An- inspiration for new ways to move will motivate peo- droid application that randomly assigns each participant ple more because varied content is important to keep to either the content-based or the user-based collabora- the users engaged [33]. As such, we hypothesize that tive filtering RS algorithm. Expecting different effects on momentary motivation to move, and thereby also the motivation and behavior, we hypothesize that collabora- amount of executed PAs and tips, will be higher in the tive filtering will provide inspiration with new ways to collaborative filtering group. move, and motivate users more than the familiar items Since both groups of participants receive an app aimed suggested by the content-based algorithm. at increasing PA, we expect that both groups will have To the best of our knowledge, the most optimal type of more PA and less SB in the post-test compared to the pre- algorithm for an HRS in the domain of PA has not been test. By following a between-subject study design, the investigated. Understanding how the algorithms sepa- long-term effect of the applied system can be assessed as rately affect motivation and behavior change is impor- a whole [12], allowing us to compare the evolution in mo- tant before combining them in a hybrid system. As such, tivation regulation style and behavior change between this study will contribute to new insights in effective the two groups. Following the SDT, the autonomous algorithms for developers of future HRSs. For example, motivation regulation types are associated with people’s future hybrid RS algorithms can assign different weights own willingness to engage in the behavior and with more to content-based and collaborative filtering recommen- psychological health, while controlled motivation is as- dation outcomes, depending on the degree to which the sociated with pressure to behave in a certain way [26]. user prefers a familiar routine or varied inspiration in Because we expect more enjoyment with the inspiring daily activities. recommendations of the collaborative filtering group [9], we hypothesize that their autonomous motivation for PA will increase. As a result of higher autonomous moti- 6 Joint Proceedings of the ACM IUI Workshops 2023, March 2023, Sydney, Australia References ing review, DIGITAL HEALTH 5 (2019) 1–19. doi:10.1177/2055207618824727. [1] World Health Organization, World health statis- [10] A. Direito, E. Carraça, J. Rawstorn, R. Whittaker, tics 2020: monitoring health for the SDGs, R. Maddison, mHealth Technologies to Influence sustainable development goals, 2020. URL: Physical Activity and Sedentary Behaviors: Behav- https://apps.who.int/iris/bitstream/handle/10665/ ior Change Techniques, Systematic Review and 332070/9789240005105-eng.pdf, oCLC: 960910741. Meta-Analysis of Randomized Controlled Trials, [2] World Health Organization, WHO guidelines Annals of Behavioral Medicine 51 (2017) 226–239. on physical activity and sedentary behaviour., doi:10.1007/s12160-016-9846-0. 2020. URL: http://www.ncbi.nlm.nih.gov/books/ [11] H. Torkamaan, J. Ziegler, Recommendations as NBK566045/, oCLC: 1237095892. Challenges: Estimating Required Effort and User [3] A. Pelliccia, S. Sharma, S. Gati, et al., 2020 ESC Ability for Health Behavior Change Recommen- Guidelines on sports cardiology and exercise in pa- dations, in: 27th International Conference on In- tients with cardiovascular disease, European Heart telligent User Interfaces, ACM, Helsinki Finland, Journal 42 (2021) 17–96. URL: https://academic. 2022, pp. 106–119. URL: https://dl.acm.org/doi/ oup.com/eurheartj/article/42/1/17/5898937. doi:10. 10.1145/3490099.3511118. doi:10.1145/3490099. 1093/eurheartj/ehaa605. 3511118. [4] World Health Organization Global Observatory [12] F. Ricci, L. Rokach, B. Shapira (Eds.), Recommender for eHealth, World Health Organization, MHealth: Systems Handbook, Springer US, New York, NY, new horizons for health through mobile tech- 2022. doi:10.1007/978-1-0716-2197-4. nologies., World Health Organization, Geneva, [13] S. Hors-Fraile, O. Rivera-Romero, F. Schneider, 2011. URL: http://www.who.int/goe/publications/ L. Fernandez-Luque, F. Luna-Perejon, A. Civit- goe_mhealth_web.pdf, oCLC: 759446797. Balcells, H. de Vries, Analyzing recommender [5] P. Liao, K. Greenewald, P. Klasnja, S. Murphy, systems for health promotion using a multidisci- Personalized HeartSteps: A Reinforcement Learn- plinary taxonomy: A scoping review, International ing Algorithm for Optimizing Physical Activity, Journal of Medical Informatics 114 (2018) 143–155. Proceedings of the ACM on Interactive, Mobile, URL: https://linkinghub.elsevier.com/retrieve/pii/ Wearable and Ubiquitous Technologies 4 (2020) S1386505617304690. doi:10.1016/j.ijmedinf. 1–22. URL: https://dl.acm.org/doi/10.1145/3381007. 2017.12.018. doi:10.1145/3381007. [14] H. Alcaraz-Herrera, J. Cartlidge, Z. Toumpakari, [6] J. P. Sanders, S. J. Biddle, K. Gokal, L. B. M. Western, I. Palomares, Evorecsys: Evo- Sherar, M. Skrybant, H. M. Parretti, N. Ives, lutionary framework for health and well-being T. Yates, N. Mutrie, A. J. Daley, ‘Snacktiv- recommender systems, User Modeling and ity™’ to increase physical activity: Time to User-Adapted Interaction (2022). doi:10.1007/ try something different?, Preventive Medicine s11257-021-09318-3. 153 (2021) 106851. URL: https://linkinghub. [15] B. Smyth, A. Lawlor, J. Berndsen, C. Feely, Rec- elsevier.com/retrieve/pii/S0091743521004205. ommendations for marathon runners: on the ap- doi:10.1016/j.ypmed.2021.106851. plication of recommender systems and machine [7] M. Rabbi, A. Pfammatter, M. Zhang, B. Spring, learning to support recreational marathon runners, T. Choudhury, Automated personalized feedback User Modeling and User-Adapted Interaction (2021) for physical activity and dietary behavior change 787–838. doi:10.1007/s11257-021-09299-3. with mobile phones: A randomized controlled trial [16] Y. Blanco-Fernandez, J. Pazos-arias, A. Gil-Solla, on adults, JMIR mHealth and uHealth 3 (2015) e42. M. Ramos-Cabrer, M. Lopez-Nores, Providing doi:10.2196/mhealth.4160. entertainment by content-based filtering and se- [8] J. Woodward, Y.-P. Chen, K. Jurczyk, K. M. Ross, mantic reasoning in intelligent recommender sys- L. Anthony, J. Ruiz, A Survey of Notification De- tems, IEEE Transactions on Consumer Electron- signs in Commercial mHealth Apps, in: Extended ics 54 (2008) 727–735. doi:10.1109/TCE.2008. Abstracts of the 2021 CHI Conference on Human 4560154. Factors in Computing Systems, ACM, Yokohama [17] R. Saia, L. Boratto, S. Carta, A new perspective on Japan, 2021, pp. 1–7. URL: https://dl.acm.org/doi/ recommender systems: A class path information 10.1145/3411763.3451639. doi:10.1145/3411763. model, in: 2015 Science and Information Confer- 3451639. ence (SAI), IEEE, London, United Kingdom, 2015, [9] K. L. Cheung, D. Durusu, X. Sui, H. de Vries, How pp. 578–585. doi:10.1109/SAI.2015.7237201. recommender systems could support and enhance [18] Y.-S. Cheng, P.-Y. Hsu, Y.-C. Liu, Identifying and rec- computer-tailored digital health programs: A scop- ommending user-interested attributes with values, 7 Joint Proceedings of the ACM IUI Workshops 2023, March 2023, Sydney, Australia Industrial Management & Data Systems 118 (2018) K. Patrick, K. J. Calfas, J. F. Sallis, Reliability and 765–781. doi:10.1108/IMDS-04-2017-0164. Validity of the Sedentary Behavior Questionnaire [19] D. Kotkov, J. Veijalainen, S. Wang, How does (SBQ) for Adults, Journal of Physical Activity and serendipity affect diversity in recommender sys- Health 7 (2010) 697–705. doi:10.1123/jpah.7.6. tems? A serendipity-oriented greedy algorithm, 697. Computing 102 (2020) 393–411. doi:10.1007/ [30] F. Wan, Statistical analysis of two arm random- s00607-018-0687-5. ized pre-post designs with one post-treatment mea- [20] P. Lally, C. H. M. van Jaarsveld, H. W. W. Potts, surement, BMC Medical Research Methodology 21 J. Wardle, How are habits formed: Modelling habit (2021) 150. doi:10.1186/s12874-021-01323-9. formation in the real world, European Journal of So- [31] B. P. Knijnenburg, M. C. Willemsen, Z. Gantner, cial Psychology 40 (2010) 998–1009. doi:10.1002/ H. Soncu, C. Newell, Explaining the user expe- ejsp.674. rience of recommender systems, User Modeling [21] B. E. Ainsworth, W. L. Haskell, S. D. Herrmann, and User-Adapted Interaction 22 (2012) 441–504. N. Meckes, D. R. Bassett, C. Tudor-Locke, J. L. Greer, doi:10.1007/s11257-011-9118-4. J. Vezina, M. C. Whitt-Glover, A. S. Leon, 2011 Com- [32] F. Zafari, I. Moser, T. Baarslag, Modelling and pendium of Physical Activities: A Second Update analysis of temporal preference drifts using a of Codes and MET Values, Medicine & Science in component-based factorised latent approach, Ex- Sports & Exercise 43 (2011) 1575–1581. URL: https: pert Systems with Applications 116 (2019) 186–208. //journals.lww.com/00005768-201108000-00025. URL: https://linkinghub.elsevier.com/retrieve/pii/ doi:10.1249/MSS.0b013e31821ece12. S0957417418305815. doi:10.1016/j.eswa.2018. [22] L. Baltrunas, X. Amatriain, Towards time- 09.010. dependant recommendation based on implicit feed- [33] J. Freyne, J. Yin, E. Brindal, G. A. Hendrie, back, 2009. S. Berkovsky, M. Noakes, Push Notifications in Diet [23] J. D. Finger, J. Tafforeau, L. Gisle, L. Oja, T. Ziese, Apps: Influencing Engagement Times and Tasks, J. Thelen, G. B. M. Mensink, C. Lange, Develop- International Journal of Human–Computer Interac- ment of the European Health Interview Survey tion 33 (2017) 833–845. doi:10.1080/10447318. - Physical Activity Questionnaire (EHIS-PAQ) to 2017.1289725. monitor physical activity in the European Union, Archives of Public Health 73 (2015) 59. doi:10. 1186/s13690-015-0110-z. [24] R. Adams, Revised Physical Activity Readi- ness Questionnaire, Canadian Family Physician Medecin De Famille Canadien 45 (1999) 992, 995, 1004–1005. [25] S. Pekár, M. Brabec, Generalized estimating equa- tions: A pragmatic and flexible approach to the marginal GLM modelling of correlated data in the behavioural sciences, Ethology 124 (2018) 86–93. doi:10.1111/eth.12713. [26] E. L. Deci, R. M. Ryan, Self-determination the- ory: A macrotheory of human motivation, develop- ment, and health., Canadian Psychology/Psycholo- gie canadienne 49 (2008) 182–185. doi:10.1037/ a0012801. [27] Deci, Edward L, Richard M. Ryan, Exercise Self- Regulation Questionnaires, Self-Determination Theory: An Approach to Human Motivation and Personality—The Self-Regulation Questionnaires (2004). [28] P. M. Wilson, W. M. Rodgers, C. C. Loitz, G. Scime, “It’s Who I Am . . . Really!’ The Importance of Inte- grated Regulation in Exercise Contexts1, Journal of Applied Biobehavioral Research 11 (2007) 79–104. doi:10.1111/j.1751-9861.2006.tb00021.x. [29] D. E. Rosenberg, G. J. Norman, N. Wagner, 8