This study employed a clustered cross-over incomplete block factorial randomization trial, a rigorous design that enables the efficient study of multiple interventions while controlling for bias. The classrooms of students in grades 5–10 at four Milwaukee schools were stratified by school and grade, then randomly assigned to one of four conditions (peak team only, Brain Agents only, both, or control) using a reproducible random-number generator. After one semester, conditions were rotated following an incomplete block design so that each classroom experienced at least one active intervention but did not repeat a condition over four semesters, 2022–24 (Appendix 1). The cross-over design ensured that each student had the opportunity to experience both interventions, reducing the impact of pre-existing differences. The incomplete block factorial structure allowed researchers to examine both individual and combined effects of the interventions. This approach minimizes selection bias, balances intervention exposure, and enhances the reliability of findings (Reich et al., 2012). This design minimized selection bias by ensuring groups were randomly assigned at the classroom level. It also balanced exposure to each condition and enhanced reliability of comparisons across interventions. This structured and rigorous approach supports internal validity and limits the influence of selection and measurement bias. The study specifically integrated EI principles within a SEL framework to assess improvements in self-awareness, emotional regulation, and social adaptability.

The peak team coach mentor program, developed by STRYV365, a non-profit organization based in Milwaukee, Wisconsin, aims to strengthen students' social-emotional skills and resilience, as shown in a prior pilot evaluation of the program documenting teacher-reported improvements in classroom behavior and student needs (Meurer et al., 2025). The program uses a trauma-informed approach that incorporates sports, mentorship, and a separate but complementary video game called Brain Agents. The peak team curriculum was designed in collaboration with educational and behavioral health experts, aligning with the principles of trauma-informed care as outlined by [Substance Abuse and Mental Health Services Administration (SAMHSA) (2014)]. The program draws on evidence-based practices and frameworks, including Communication, Offering, Positivity, Empowerment (COPE) and Catch yourself, Relax, Assess, Focus, Think (CRAFT) approaches, which focus on communication, empowerment, and decision-making [American Psychological Association, 2011; The Center for Adolescent Substance Use Research, 2018]. Grounded in SAMHSA's six principles of trauma-informed care—safety, trust, peer support, collaboration, empowerment, and cultural sensitivity—the peak team program helps students to develop self-awareness, self-management, and relationship and decision-making skills, providing them with the tools to overcome adversity and thrive. Peak team fosters emotional regulation, communication, and teamwork in real-world settings. The program's name, peak team, is lower case to show that both the youth participants and the coaches are on the same power dynamic level.

Peak team and Brain Agents aim to target emotional regulation, relationship building, responsible decision-making, and social awareness through social-emotional training. They impact mental health outcomes through reduced symptoms, protective buffering, and academic synergy (Babad et al., 2022). The STRYV365 peak team program and Brain Agents video game align with Mayer and Salovey's model by integrating activities and game-based learning that target EI domains (Table 1). For example, the physical activities and mentorship sessions cultivate self-awareness and social awareness, while the video game fosters emotional regulation and problem-solving skills, enhancing self-management and responsible decision-making.

The peak team program incorporates play, sports, and games into educational settings to enhance student engagement and learning. Peak team incorporates physical activities such as basketball, soccer, and team-based games, which are designed to develop EI competencies like emotional regulation, resilience, and teamwork. These activities intentionally promote emotional growth and social skills alongside physical engagement. Peak team differs from regular gym classes by integrating trauma-informed coaching, SEL, and mentorship into sports activities, focusing on emotional regulation, resilience, and teamwork rather than just physical fitness. Unlike traditional gym lessons, it includes structured reflections, coping strategies, and guided discussions, using sports as a vehicle for holistic student development. Physical activity and sports promote teamwork, discipline, and emotional regulation, while game-based learning makes education more appealing and effective. These components are particularly valuable in SEL programs, where engaging activities reinforce the skills being taught and offer practical opportunities for students to apply lessons learned (Granic et al., 2014).

The peak team program was provided from four to ten weeks in gym or physical education classes with a range of 9–12 sessions each including 4–11 h of students and coach interaction time. Students participated in peak team sessions once weekly during the assigned semesters. Each session lasted approximately 30–60 min, featuring activities such as basketball, soccer, games, artwork, reflective writing, and discussion to develop their SEL skills. The delivery of peak team sports, games, and art activities following the CRAFT framework was standardized and consistent across the four schools and five grades over two years. Coaches were flexible in responding to specific school and class needs and available resources with adjustments and accommodations made based on grade level and individual student's ability.

The Brain Agents video game, developed in collaboration with behavioral health professionals and game designers from Night City and Columbia College Chicago, targeted students aged 10–15 years (Figure 1 screenshot of title). The game integrates SEL objectives such as trauma resilience, coping strategies, cognitive reframing, and emotional awareness within an engaging digital format. Brain Agents includes cognitive reframing, emotional regulation exercises, problem-solving challenges, and interactive social scenarios to help students recognize emotions, manage stress, and build resilience. These elements can improve mental health symptoms by enhancing coping strategies for anxiety and depression, promoting self-awareness and impulse control, and reinforcing positive decision-making skills through engaging, low-risk digital simulations (Fleming et al., 2017).

Players of the game, shown as explorers on a space vessel, respond to a distress signal and encounter an alien computer program, MAL, which infects both computer systems and human minds, causing cognitive distortions (Figure 2 screenshot of gameplay showing narrative content). The game features five mini-games designed to help players practice coping mechanisms, emotional awareness, and social connection with non-player characters. These mini-games include:

The game's development used the Unity game engine to ensure compatibility with various devices, including Chromebooks. An early access model allowed students to interact with the game while it was still in development, providing feedback that shaped continuous improvements. Based on student feedback in year one, the Brain Agents game optimized performance and graphics, clarified mini-game missions, allowed for characters customization, and characters were given a tactical roll in year two. Students played Brain Agents two to three days weekly for four to five weeks, averaging ten sessions per semester and 23 min per session, totalling approximately 3.8 h of playtime. Most students did not complete the entire game. Students completed an average three of 15 total missions. Brain Agents' play time, outcomes, and in-depth qualitative analysis of its impact on students' SEL competencies is described elsewhere (Liverman et al., 2025).

The Medical College of Wisconsin Human Rights Review Board approved the research protocol.

STRYV365 partnered with four Milwaukee schools—two public and two private choice—serving a total of 1,626 students in grades 5–10. The demographic composition included 61% Black/African American, 19% White, 11% Hispanic/Latinx, 6% Asian/Hmong, and 3% mixed racial/ethnic backgrounds. Approximately 69% of students experienced economic disadvantage, and academic performance varied, with 58% scoring below basic on standardized exams. Despite these challenges, 88% of students at these schools earn high school diplomas, indicating a strong commitment to education and baseline resilience among the student population.

A power analysis suggested that to detect a small to moderate effect size (Cohen's d = 0.3) for self-awareness, relationship skills, coping options, and emotional regulation, with 80% power (β = 0.20) and a significance level of 0.05, each of the four intervention groups (peak team, Brain Agents, both, control) would require approximately 175 participants per group. This results in a total sample size of approximately 702 participants to ensure adequate statistical power for detecting meaningful program effects.

Among the 1,626 eligible students, parents/primary caregivers of 399 students consented to their child participating in the research program evaluation, and 329 students assented to participation in the study. Students were tracked through six survey points and qualitative evaluations to assess the impact of the interventions.

Data collection included a 41-question survey instrument in REDCap (Harris et al., 2019; see Appendix 2) derived from validated and reliable items provided by the Life Paths Research Center (Hamby et al., 2018) and aligned with the competencies outlined by the Collaborative for Academic, Social, and Emotional Learning [Collaborative for Academic, Social and Emotional Learning (CASEL), 2025]. The surveys administered in the fall, winter, and spring of each intervention year assessed key SEL domains such as self-awareness, self-management, relationship skills, responsible decision-making, social awareness, and adversity. The surveys also included recent feelings of depression and anxiety, using items from the Patient Health Questionnaire-2 (Richardson et al., 2010) and Generalized Anxiety Disorder-7 questionnaires (Mossman et al., 2017). Students who reported daily symptoms of anxiety or depression were promptly referred to their school behavioral health counselor and a child psychologist through STRYV365.

Participant survey fatigue was reduced by limiting gender and race/ethnicity questions to baseline only. Distinct post-intervention questions for each program reduced overlap and improved attribution of effects to specific interventions. Students answered 15 core questions every cycle. After peak team, they answered 13 multiple choice questions and provided two open-response comments. After Brain Agents, they answered nine questions and provided two open-response comments. Three cyberbullying questions were asked in year two only. Survey items were not rotated to ensure consistent data collection across the four group assignments. Among the 329 assenting students, 99 (30%) completed all six surveys, 87 (26%) completed five surveys, 73 (22%) completed four, 32 (10%) did three, 24 (7%) did two, 11 (3%) did one, and three students (1%) did no surveys.

In addition to surveys, qualitative data were collected through focus groups and interviews, conducted at the end of the fall and spring semesters. Focus groups with 5th−7th graders encouraged peer interaction, while individual interviews with 8th−9th graders (and tenth in year two) allowed for more in-depth discussions. The qualitative evaluation was guided by a semi-structured interview protocol (see Appendix 3), ensuring that both open-ended responses and targeted questions were captured.

The study employed a rigorous, equity-centered qualitative methodology grounded in sociocultural theory and guided by SEL and EI frameworks. Reflexive co-construction, diverse researcher positionalities, thematic coding, and mixed human-AI review processes contributed to a methodologically robust and deeply contextualized understanding of how trauma-informed interventions affect student emotional and social development.

Transcripts of focus groups and interviews were professionally transcribed and stored securely. Thematic analysis was conducted using Dedoose software (Dedoose, 2021), with ten trained coders employing an iterative process of generating and refining codes and themes with a focus on the CASEL framework (2023). Two independent coders reviewed each transcript and discussed discrepancies to optimize intercoder reliability and to minimize subjective bias. This approach ensured a comprehensive understanding of the qualitative data, with a focus on identifying patterns related to the impact of the interventions on students' SEL outcomes.

With 23,930 words of deidentified quotes, we prompted ChatGPT 4.0 (OpenAI, 2024) to “select and report relevant, compelling, and representative quotes for scientific publication.” While ChatGPT was used to assist in organizing and presenting representative quotes, all qualitative analyses, including theme development and coding, were conducted by a trained team of coders using an established thematic analysis approach. ChatGPT was used solely as a tool to help select quotes after themes were identified, enhancing clarity and accessibility of findings without replacing human analysis or interpretation. Seven human coders reviewed, selected, and confirmed top quotes per theme which were generated by ChatGPT.

A series of general linear models were conducted to evaluate the effects of the STRYV365 interventions peak team and Brain Agents on clustered student surveyed outcomes. The models examined both main effects of each intervention and their interaction, adjusting for contextual and demographic covariates. Primary outcomes included relationship quality, coping strategies, and emotional regulation assessed via student-reported survey items. Self-awareness was examined as a secondary outcome because baseline assessment was relatively high.

Independent variables included participation in peak team and Brain Agents (both binary), time of assessment (semester, categorical), and grade-level classroom clusters (16 categories). Covariates also included indicators of family adversity: having a family member who had died, been incarcerated, or used substances of abuse (binary variables). Clustered baseline scores for each outcome (relationship quality) were included as covariates. The models incorporated weights accounting for the number of students in each cluster.

Both Type I (sequential) and Type III (adjusted) sums of squares were examined to evaluate the contribution of each predictor to the models. Model fit was assessed using R² values (shown without a carry-over effect). SAS software (version 9.4) was used for all analyses. Using the Bonferroni correction to control for an increase in false positives when conducting multiple comparisons on the same dataset, statistical significance was set at p < 0.017 due to three primary outcomes of relationship quality, coping strategies, and emotional regulation (self-awareness was excluded because it was generally high at baseline), with marginal effects (p < 0.05) noted.

Tables 7, 8 shows persistently high rates of adversity, including death of a close person (69–74%) and family incarceration (50–56%), with modest improvements in depression and anxiety symptoms over time. These adversity factors had measurable influence in regression modeling, particularly in reducing self-awareness and relationship skills.

The regression models yielded high R² values (0.65–0.88) in Table 3. This is expected in studies where baseline values of each SEL domain are included as covariates, because SEL constructs show strong temporal stability. When baseline scores account for a majority of variance, the remaining variance attributed to interventions or adversity factors becomes easier to interpret, but model fit necessarily appears high. In this study, high R² values primarily reflected baseline SEL score covariates, cluster-level effects (classroom assignment), and time effects reflecting semester-to-semester developmental changes. Thus, high R² values do not indicate overfitting but rather the appropriate inclusion of baseline SEL scores and cluster controls required by the factorial randomized crossover design.

The most statistically significant finding (p = 0.0003) is the profound negative impact of family incarceration on students' self-awareness. For interventions, we found strong, convergent evidence (both qualitative and quantitative, p = 0.04) that the labor-intensive peak team mentorship model successfully improved emotional regulation.

The findings regarding the Brain Agents digital intervention are more complex and present a paradox. While the regression model identified a preliminary statistical relationship with improved coping strategies (p = 0.03), implementation data showed that most students did not engage deeply with the program, with an average of three completed missions out of 15 available. However, qualitative analyses from our companion manuscript demonstrated that even brief exposure to the early Brain Agents missions helped students practice positive self-talk, breathing techniques, emotional naming, and cognitive reframing, skills students explicitly linked to calming down and managing stress (Liverman et al., 2025). Teacher perspectives further corroborated that Brain Agents increased students' emotional vocabulary, willingness to pause before reacting, and ability to articulate stressors (Stoltenburg et al., 2024). These findings clarify why limited gameplay may still translate into measurable improvements: the earliest mini-games contain core SEL mechanics that students readily applied to real-life situations. At the same time, inconsistent school-day scheduling, device access, and the early-access stage of game development limited sustained engagement and likely constrained the full potential of the intervention. Overall, across qualitative and quantitative sources, Brain Agents appears to serve as a concise, accessible SEL tool that can introduce and rehearse coping strategies, while peak team provides extended, relational opportunities for deeper practice and reinforcement.

The results of this study demonstrate measurable improvements across key domains of Mayer and Salovey's EI framework. The peak team program's activities supported students' ability to manage emotions and foster relationships, aligning with managing emotions and social skills components of the framework. Similarly, the Brain Agents game promoted emotional regulation and critical thinking, reflecting the understanding of emotional and problem-solving components of EI. These findings underscore the effectiveness of integrating EI-focused activities into trauma-informed interventions for children facing adversity.

The statistical results further validate the qualitative impressions: peak team was significantly associated with gains in emotional regulation, while Brain Agents promoted coping strategies—a core element of self-management in the EI framework. These results strengthen the case for combining mentorship and game-based SEL tools. The regression models explained a substantial portion of variance (R² ranging from 0.65 to 0.88), especially for relationship and regulation outcomes.

The findings from this study are consistent with previous research on the effectiveness of SEL programs. For example, Durlak et al. (2011) demonstrated that SEL programs can significantly enhance students' social-emotional competencies, leading to improved academic performance and peer relationships. The innovative aspect of the STRYV365 program lies in its integration of the Brain Agents video game, which add interactive and engaging elements to the traditional SEL curriculum. Educational video games have been shown to reinforce social-emotional skills and engage students in ways that traditional methods may not (Toh and Kirschner, 2023). The combination of digital and novel coach mentoring methods by STRYV365 offers a unique and effective approach to SEL, particularly for children facing adversity. These findings contribute to the growing body of evidence supporting the integration of EI into educational interventions. Programs offered by STRYV365 complement traditional SEL curricula by incorporating novel methods such as mentorship and game-based learning, which uniquely address emotional regulation and social engagement challenges among students facing adversity.

Given the demographic composition of the study population, this research underscores the importance of culturally responsive approaches to EI development. The STRYV365 peak team coaches offered diverse sociodemographic backgrounds and experiences. Future studies should explore how tailored interventions might address specific cultural or socio-economic factors affecting emotional and social development. Building on this foundation, the STRYV365 model could be adapted for other sociocultural contexts by engaging local educators, community mentors, and families to align program content with regional values and experiences. In suburban or rural settings, or in communities with different racial and ethnic compositions, implementation might emphasize contextual strengths such as intergenerational mentoring, nature-based learning, and digital accessibility. Cross-site pilot studies in diverse geographic and cultural environments would help identify which trauma-informed and EI components are most universally transferable and which require local tailoring.

This study has several limitations. The attendance and academic performance of participating students is not known since schools did not share this data with researchers. Self-reported data may introduce bias, as students might under- or overstate their emotions and behaviors. Survey fatigue occurred due to long surveys, particularly when students were exposed to both interventions. Varying intervention duration (4–10 weeks) might have affected consistency in program exposure. Most students did not complete the full Brain Agents game, likely limiting its full effect. Some students may benefit from more structured guidance on how to transfer skills learned in the game to real-life situations. The clustered cross-over design, while rigorous, may limit generalizability to non-urban or differently structured schools. Limitations of the clustering include a small sample size with 16 clusters measured four times. While regression models controlled for adversity and clustering, the small number of clusters and survey attrition limit generalizability. The models identified some effects as marginal, warranting replication in larger samples. Artificial intelligence, optimized for fluency and pattern recognition, may have systematically selected the most fluent or dramatic quotes rather than the most typical or common ones. This can skew data representation, overemphasize certain themes, and overlook negative cases. External influences, such as school environment and family context, may confound the results. However, using at least two coders per theme to check artificial intelligence quote selections mitigated this limitation.

This study suggests that streamlining surveys by asking fewer key questions more frequently could be used in the future. Strategies should be investigated to increase engagement and completion rates for digital tools like Brain Agents. The STRYV365 model should be expanded to other school settings to test scalability and adaptability. Culturally tailored versions of the program should be explored to enhance relevance across diverse populations. Further research is needed on the long-term impacts of SEL interventions.

The peak team and Brain Agents interventions offer a scalable model for enhancing SEL through mentorship and game-based learning. For schools, the most immediate and impactful recommendation is to invest in coaches and mentors trained in trauma-informed care, as intensive relational support was the primary driver of improved emotional regulation in this study. Schools can integrate peak team-style coaching with structured physical activities, and reflective practices into physical education, advisory periods, or after-school programs to ensure consistent, relationship-centered SEL delivery. For districts adopting digital tools, Brain Agents demonstrates that even brief exposure to well-designed SEL game mechanics can strengthen coping and emotional awareness, but sustained engagement requires predictable scheduling, teacher support, and reliable device access.

For game developers, these findings highlight the importance of designing compelling early-game experiences, shorter mission structures, culturally relevant characters, and features that encourage replay and completion of missions. Developers should prioritize usability testing in real classroom environments and collaborate closely with educators to ensure alignment with school-day routines.

These programs can be adapted for high-need schools using existing technology, embedded mentorship, and ongoing professional development. Sustainability can be strengthened through community partnerships, teacher training, and iterative improvements to both coach-based and digital components. Scaling these models in diverse educational settings has the potential to meaningfully enhance student resilience, emotional well-being, and academic engagement.

The positive outcomes associated with the STRYV365 program underscore the importance of integrating SEL curricula with innovative tools like educational video games to enhance engagement and learning (Greenberg et al., 2003). These findings align with broader research emphasizing the need for comprehensive, innovative approaches to SEL that can be adapted to diverse learning environments (Greenberg et al., 2003; Taylor et al., 2017). The success of the peak team program highlights the value of mentorship and coaching in fostering resilience and social-emotional development. These results suggest that similar programs could be effectively implemented in other school settings to address the needs of students exposed to ACEs. Policymakers and educators should consider these approaches as part of broader efforts to create supportive and inclusive learning environments.