School-based high-intensity interval training improves classroom engagement more than moderate-intensity continuous training in children aged 8–9 years: a randomized controlled trial

Technical Summary: School-Based HIIT vs. MICT on Classroom Engagement

Problem Statement
Classroom engagement is a critical behavioral predictor of academic achievement and school functioning in children. While school-based exercise interventions are known to potentially improve cognitive and behavioral outcomes, there is limited evidence directly comparing the effects of High-Intensity Interval Training (HIIT) and Moderate-Intensity Continuous Training (MICT) on classroom engagement. Furthermore, traditional assessment methods (e.g., teacher reports or standardized cognitive tests) often fail to capture dynamic, real-time attentional behaviors in authentic classroom settings. This study addresses the gap in understanding how different exercise intensities and modalities affect observable engagement behaviors in children aged 8–9 years, utilizing an objective, computer vision-based assessment system.

Methodology
The study was a three-arm parallel-group randomized controlled trial (RCT) conducted with 72 third-grade students (aged 8–9 years) from a single primary school. Participants were randomly assigned (1:1:1) to one of three groups:

1. HIIT Group: Performed 15-minute sessions (3 times/week for 6 weeks) involving whole-body exercises (jumping jacks, scissor jumps, vertical jumps, burpees) in a 1:1 work-to-rest ratio, maintaining an intensity of ≥85% of predicted maximal heart rate (HRmax).
2. MICT Group: Performed 15-minute sessions (3 times/week for 6 weeks) of continuous aerobic running (circular, curved, and shuttle running) at 60–69% HRmax.
3. Control Group: Continued routine school activities and physical education without additional structured exercise.

Intervention and Monitoring:

• Duration: 6 weeks, 18 total sessions.
• Intensity Verification: Heart rate was monitored using Polar devices to ensure adherence to target zones (HIIT: ≥171 bpm; MICT: 121–139 bpm).
• Outcome Assessment: Classroom engagement was measured immediately post-intervention during regular core academic lessons (Chinese, Mathematics, English).
• Measurement Technology: An Intel RealSense 3D camera system was employed to non-invasively capture depth video streams. Using the RealSense SDK (Pose Detection, Gaze Tracking, Face Detection, Landmark Detection), four behavioral indicators were extracted:
  1. Head posture
  2. Gaze direction
  3. Eye openness
  4. Facial orientation
• Metrics: A composite engagement score was calculated as the geometric mean of the four standardized indicators. "Sustained engagement duration" was defined as the proportion of class time where the composite score was ≥0.60.
• Statistical Analysis: Linear mixed-effects models (with participant ID as a random intercept) were used to test for Group × Time interactions, followed by Bonferroni-adjusted post hoc comparisons.

Key Results

• Composite Engagement: A significant Group × Time interaction was observed ($p < 0.01$, partial $\eta^2 = 0.163$). Both HIIT and MICT groups showed significant improvements from baseline to post-intervention ($p < 0.01$). However, the HIIT group demonstrated significantly greater post-intervention composite engagement scores compared to both the MICT and control groups ($p < 0.01$).
• Behavioral Indicators: Significant Group × Time interactions were found for all four indicators (head posture, gaze direction, eye openness, facial orientation; all $p < 0.01$).
  • HIIT Superiority: The HIIT group showed significantly higher post-intervention scores than the MICT and control groups in head posture, gaze direction, and facial orientation.
  • Eye Openness: While both exercise groups improved significantly from baseline, there was no significant difference between HIIT and MICT for eye openness at post-intervention.
• Sustained Engagement Duration: A significant Group × Time interaction was observed ($p < 0.01$, partial $\eta^2 = 0.129$). The duration of sustained engagement increased significantly only in the HIIT group ($p < 0.01$). Post-intervention, the HIIT group had significantly longer sustained engagement than the control group, while the MICT group did not differ significantly from the control.

Key Contributions

1. Comparative Efficacy: The study provides empirical evidence that while both HIIT and MICT improve classroom engagement, HIIT yields superior results across multiple behavioral indices, including the composite score and sustained engagement duration.
2. Objective Assessment in Natural Settings: The research demonstrates the feasibility and utility of using Intel RealSense-based computer vision to objectively quantify classroom engagement behaviors in authentic school environments, moving beyond subjective teacher ratings.
3. Protocol Specificity: By matching duration and frequency but varying intensity and movement characteristics (whole-body intervals vs. continuous running), the study isolates the combined effects of high intensity and movement complexity on behavioral outcomes.

Significance and Claims
The authors conclude that brief, school-based exercise is a feasible strategy for enhancing behavioral engagement during classroom learning. Specifically, HIIT appears to offer additional benefits over MICT for young children (aged 8–9). The study posits that the superior effects of HIIT likely reflect a combined influence of higher exercise intensity, greater motor complexity, and increased cognitive engagement, rather than intensity alone.

The paper modestly frames its findings as evidence of changes in observable classroom behaviors rather than definitive improvements in underlying cognitive attention or executive function, noting that standardized cognitive tests were not included. The authors emphasize that these results support the integration of brief, high-intensity activity into school schedules to support learning readiness, while calling for future multi-site trials incorporating cognitive and academic outcomes to clarify mechanisms and generalizability.