HelperFriend was developed by a multidisciplinary team that included nutritionists, psychologists, physical activity experts, human–computer interaction experts, and software engineers based on published design methodology [43]. The methodology included activities from game implementation to evaluation based on 4 essential principles: a procedure-centric approach, expert collaboration, agile development, and low-cost modeling. The knowledge domains of HelperFriend are physical activity, healthy eating, and socioemotional wellness.

HelperFriend integrates experiential and vicarious learning. In experiential learning environment, learners engage in direct experiences to enhance their knowledge, skills, and values through human–environment interaction in a cycle of doing, reflecting, concluding, and trying the learned experience [44]. In vicarious learning, individuals learn from the experiences of others (eg, by observing the choices another person makes and the consequences they have on their health). By observing the behavior of others, individuals can identify difficulties and expectations associated with behaviors and acquire the information and competencies to perform the behavior successfully [45].

In addition, several behavior change techniques [46] were integrated into the game elements to generate an attractive and stimulating environment in which knowledge and healthy lifestyle behaviors are encouraged and reinforced: instruction on how to perform the behavior, providing information about health consequences, behavioral practice, behavioral substitution, incentives and rewards, goal setting, reviewing behaviors goal, monitoring behaviors, providing feedback on behavior, discrepancies between current behaviors and goals, monitoring emotion consequences, and prompts or cues. These behavior change techniques are based on behavioral, cognitive, and social cognitive theories that have ample empirical evidence to demonstrate their usefulness in adopting healthy lifestyles [47-50].

In HelperFriend, the players are secret agents who need to care for a group of children who forgot healthy lifestyle behaviors because a villain chef erased their memory. In each match, the player needs to ensure that one of these children engages in physical activity, eats well, and performs socioemotional activities to improve their health (Figure 1). The children characters continuously interact with the player, expressing their necessities or stating situations for which they need help. Player actions that improve the children's lifestyle behaviors add points. The game session is finished when the player presses the finish button. At the end, the player can earn extra points if their decisions helped the child to meet healthy lifestyle recommendations. The player has full-body interaction to encourage physical activity and improve satisfaction and fun [51]. A video of the game is provided as Multimedia Appendix 1.

We conducted a parallel randomized controlled pilot trial over 4 weeks between May and June 2019 in an elementary school in Mexico.

School administrators and teachers gave written permission for the trial to be performed at school facilities. All study procedures were approved by the institutional review board of the Centro de Investigación Científica y de Educación Superior de Ensenada (2S.3.1 HUM 2019). No changes occurred to the methods after the beginning of the trial.

Students (n=40) from 3 school groups was considered for this study. Inclusion criteria were being aged 8 to 11 years and not receiving pharmacological treatment. Exclusion criteria were having been diagnosed with or having an ongoing neuropsychiatric disorder, a physical problem (because the game required children to interact through whole-body movements), and obesity treatment in the past 6 months. Written informed consent was obtained from parents of children who expressed interest in participating in the study.

Children were randomly allocated to either the control group or the intervention group. The children in the intervention group played HelperFriend during six 30-minute game sessions. All playing sessions were conducted over 21 days. We set up 3 gaming stations in a room; each station contained a PC, a 50-inch screen, a Kinect sensor V2, and the HelperFriend video game. Participants in the control group received only a 45-minute talk about the importance of healthy behaviors, such as engaging in physical activity, eating healthy, and maintaining socioemotional health; no further intervention was applied.

Outcomes were assessed in both groups the week after being assigned to the groups (T0) and 4 weeks after baseline (T1). The primary outcome was the gain in knowledge measured using a questionnaire (developed by the research group and designed specifically for the serious game). The questionnaire was evaluated in a pilot with 5 children and adapted. The final questionnaire consisted of 82 questions in 3 sections: physical activity (13 questions, each with 3 response options), healthy eating (64 questions, each with 3 to 5 response options in food groups, food portions equivalence, correct diet, and healthy/unhealthy food subsections), and socioemotional wellness (5 questions, each with 4 response options). Figure 5 provides some examples of the questions. Children completed the questionnaire by themselves. The sum of questions that had been appropriately answered for each section was calculated.

Secondary outcomes were intention to conduct healthy behaviors, dietary intake, and player experience satisfaction.

Children’s intention to conduct healthy behaviors was measured using a questionnaire tailored specifically for the serious game. The questionnaire was pilot-tested with 5 children and adapted. The final questionnaire (Multimedia Appendix 2) consisted of 33 questions in 3 sections: physical activity (4 questions), healthy eating and correct diet (24 questions), and socioemotional wellness (5 questions). These questions state everyday situations that children experiment in their daily lives, and they have to decide what action to take to solve the case. Physical activity, healthy eating, and socioemotional wellness questions have 3, 3, and 4 response options, respectively. Questions related to correct diet have 3 response options and use a graphical representation to facilitate children's answers (a previous study [53] used similar graphical questionnaires with children). For each question, there was only 1 appropriate answer, and children completed the questionnaire by themselves. The sum of questions that had been appropriately answered for each section was calculated.

Dietary behavior was measured with a food frequency intake questionnaire [41]. This questionnaire was explicitly designed for the diet of school-age Mexican children and included features of the highest validated food frequency questionnaires. It consists of 78 food items considered to be indicators for healthy and unhealthy eating behaviors. For each item, a 7-point scale, from 0 (never) to 6 (two or more times per day), is used to indicate the frequency that the food is consumed. Each question was scored individually. In order to facilitate the completion of the questionnaire, a facilitator read the questions to the children, who only had to answer with the number of times that they had eaten the food in the past month.

Player experience satisfaction was only assessed in the intervention group (at T1). An adapted version of the Game User Experience Satisfaction Scale [54] was used. The scale consists of 23 questions in 7 domains: playability, narratives, enjoyment, personal gratification, creative freedom, audio aesthetics, and visual aesthetics. We asked participants to specify their agreement level using a 5-point Likert scale, from 1 (totally disagree) to 5 (totally agree). A score was calculated for each domain.

Data were analyzed using SPSS software (version 26; IBM Corp). The statistical significance for all analyses was P<.05. Variables are reported as means (with standard deviations) or medians (with interquartile ranges). The normality distribution of interval variables was tested using the Shapiro-Wilk test. For metric data, differences between pretest and posttest were analyzed using 1-tailed paired t tests. For nonmetric data, the differences between pretest and posttest were analyzed using Wilcoxon signed-rank sum tests. Differences between groups were analyzed using 1-tailed independent t tests. The relationships between subscale items were tested using Cronbach α. No power estimation was performed since this was a pilot study.

Intention to perform healthy behaviors in the intervention group increased from T0 to T1 for physical activity (t₁₄=2.82, P=.006), healthy eating (t₁₄=3.44, P=.002), and socioemotional wellness (t₁₄=2.65, P=.009). There were no significant differences in intention to conduct healthy lifestyle behaviors for control group participants for physical activity (t₁₁=–0.80, P=.22), healthy eating (t₁₁=0.40, P=.40), and socioemotional wellness (t₁₁=0.23, P=.41). In addition, differences in T1 intention scores between the intervention group and control group were statistically significant for physical activity (t₂₅=1.95, P=.03), healthy eating (t₂₅=1.91, P=.03), and socioemotional wellness (t₂₅=2.43, P=.01); the intervention group scored higher than the control group for all 3 (Table 1).

Participants in the intervention group reported reduced consumption frequency of ham, sausage, soft drinks, wheat burritos, hamburgers, breaded chicken, sopes, tamales, salt peanuts, sweet cookies, potatoes chips, cake, and sweet soft cakes. Participants in the control group indicated reduced self-reported frequency intake of 5 healthy foods (cantaloupe, carrot, fish soup, fish ceviche, and fresh fruit juice) and 1 unhealthy food (bottled fruit juice) (Table 2).

Satisfaction ratings were significantly higher than the neutral value for all domains: playability (t₁₄=7.04, P<.001), narrative (t₁₄=4.00, P<.001), enjoyment (t₁₄=4.77, P<.001), creative freedom (t₁₄=7.69, P<.001), audio aesthetics (t₁₄=4.33, P<.001), personal gratification (t₁₄=5.99, P<.001), and visual aesthetics (t₁₄=5.12, P<.001). Most participants agreed that the game was easy to learn to play and use (14/14, 100%), has a clear history (12/15, 80%), is fun and original (11/15, 73%), has good music (12/15, 80%), has good graphics (12/15, 80%), and made them feel successful when they overcame the game's challenges (13/15, 87%). In addition, most participants wanted to play HelperFriend again (12/15, 80%). All measures obtained Cronbach α values ≥.73, except for narrative (Cronbach α=.56) (Table 3).

Children in the intervention group significantly improved their knowledge about physical activity (t₁₄=2.01, P=.03), healthy eating (t₁₄=3.14, P=.003), and socioemotional wellness (t₁₄=2.75, P=.008) after gameplay; thus, the first hypothesis was verified. Previous studies have also shown that video games for health can help children improve their understanding of physical activity [55] and healthy eating [35,37-39,41,55]. Other than in a recent study [42], in which stress and stress-coping strategies were included, emotional and social issues related to adopting healthy lifestyle behaviors have not been previously taken into account in interventions with serious games that have targeted healthy eating. We also identified that it is possible to teach this type of issue with serious games. Our results support those of a previous study [55], showing that it is feasible to improve physical activity and healthy eating knowledge together. These promising results may be explained by the experiential and vicarious learning environment of the game in which players observe the behavior of others [45] and engage in direct experiences through doing, reflecting, concluding, and trying the learned experience [44]. Conversely, because improving health knowledge through serious games does not necessarily result in behavioral change [28], we also plan to adjust content and learning strategies in a future version.

Second, we hypothesized that the intention to conduct physical activity, healthy eating, and healthy socioemotional behaviors would be higher after the intervention; we also verified this hypothesis. A previous study [40] also showed that video games for health could help children improve dietary and exercise attitudes, but we did not identify any studies on intention to conduct healthy socioemotional behaviors. Improved intention to engage in healthy behaviors may be as a result of integrating multiple behavior change techniques—a cornerstone for efficacy in behavior change interventions [56]. Because a change in intention leads to a small to medium change in behavior [57], these findings help assess the potential impact of video games on children's lifestyle behaviors. The smallest increase in intention was for socioemotional wellness, possibly because the intervention group already scored higher at baseline, which may be associated with the fact that the questionnaire contained simple questions that make it easy to obtain a high score.

The third hypothesis stated that, after the intervention, children's diets would improve. A lower intake frequency was found for 13 unhealthy foods (such as soft drinks, hamburgers, sweet cookies, potatoes chips, and sweet soft cakes). These changes are relevant because Mexican children commonly consume these foods in schools and at home [58] and changes in children's diets and eating habits can promote changes in the whole family [59]. Our findings confirm that games to improve healthy food consumption are beneficial, which has been demonstrated in some earlier studies (eg, [36,39,41]), but not others (eg, [42]). Unlike some previous studies (eg, [35,39]), we only achieved a reduction in the consumption of unhealthy foods. One possible explanation is that providing information or visual images of foods alone is insufficient to increase children's preferences for the intake of healthy foods [60,61]. Instead, repeated exposure to healthy foods is more effective for improving children's preferences [61]—even more so than strategies based on rewards [62]. A future version of HelperFriend should include behavior change techniques (eg, self-monitoring, setting and examining goals, and action planning) that directly support and encourage healthy food intake. Surprisingly, the participants in the control group indicated reduced intake of 5 healthy foods (cantaloupe, carrot, fish soup, fish ceviche, fresh fruit juice) and 1 unhealthy food (bottled fruit juice); however, we did not collect any other information from the control group that could facilitate the interpretation of this result.

The fourth hypothesis was also verified; children felt good during gameplay, and game acceptance was high. HelperFriend obtained very positive results on personal gratification, playability, creative freedom, enjoyment, narrative, and visual and audio aesthetics—factors which have been shown to be correlated with and predictors of learning [63]. However, for four specific aspects, there is room for improvement: (1) improving the fit of the difficulty curve of the game to the capacities of the children, (2) simplifying the game flow to foster player autonomy, (3) increasing socioemotional elements, and (4) implementing a daily activity tracking system in the game to make it easier for children to understand the daily activities they have to carry out to have healthy lifestyles. These characteristics could be essential aspects that positively influence the general perception, acceptability, and effectiveness of the game.