Actionable companion · Assessment & reasoning

Learning with agency: School-based predictive contributors to six-minute walk test performance across pubertal stages: an explainable machine learning study in Tunisian adolescents

AI-assisted editorial implementation guide, separate from the original research. Proposed actions are not claims or recommendations attributed to the source authors. No independent peer review of this guide is asserted.

Read original paper Search paper agent CC BY 4.0

Evidence and context

Title and abstract screening; stage labels describe reported settings, not verified numerical ages. Design labels require full-text appraisal. Reported stage(s): Secondary school. Verify the participant population and task before transferring the findings.

Original authors’ abstract, reproduced under the linked source license:

Abstract Background The six-minute walk test (6MWT) provides a practical measure of submaximal functional exercise capacity in adolescents and reflects the integrated response of cardiovascular, respiratory, and musculoskeletal systems during walking. Traditional age-based linear models inadequately capture nonlinear, maturation-dependent interactions among anthropometric, behavioral, and physiological factors. This study aimed to (i) identify stage-specific predictive contributors to 6MWT performance using supervised machine learning, (ii) improve interpretability through explainable artificial intelligence, and (iii) explore data-driven functional and cardiovascular profiles across peak height velocity (PHV) stages. Methods In this cross-sectional school-based study, 3,166 Tunisian adolescents (11–18 years) were categorized as pre-PHV (n = 959), around-PHV (n = 953), or post-PHV (n = 1,254). Submaximal functional exercise capacity was assessed using the 6MWT distance. Random forest regression models were developed globally and stratified by maturation stage. Predictive contributors were interpreted using permutation importance and Shapley Additive Explanations (SHAP). Exploratory k-means clustering based only on 6MWT and cardiovascular response variables was used to identify data-driven profiles within each PHV stage. Results The global model explained 40% of the variance in 6MWT distance (R2 = 0.399; RMSE = 121 m). Predictive performance varied markedly by maturation stage, with low explanatory power pre-PHV (R2 = 0.039), intermediate performance around-PHV (R2 = 0.178), and higher predictability post-PHV (R2 = 0.347). Physical activity was the highest-ranking predictor globally and across all PHV stages, followed by chronological age, particularly around-PHV. SHAP analyses showed progressively larger mean absolute contributions of physical activity across PHV stages, reaching approximately 45 m in post-PHV participants. Although the gap statistic favored a one-cluster solution, exploratory global clustering with k = 2 described a lower-response profile [6MWT distance = 490.4 ± 132.1 m; exercise-induced heart rate response (ΔHR) = 68.4 ± 12.6 bpm] and a higher-response profile (6MWT distance = 599.0 ± 157.1 m; ΔHR = 91.3 ± 13.3 bpm), both of which were represented across all PHV stages. Conclusions In this cross-sectional cohort, 6MWT performance showed maturation-stage-specific prediction patterns, but model performance remained modest, particularly before PHV. Physical activity had the strongest model contribution after PHV and may represent an important modifiable correlate, although causal inference cannot be made from these data. The exploratory functional and cardiovascular profiles should be interpreted as data-driven clusters rather than stable biological states. These findings support maturation-sensitive assessment of adolescent 6MWT performance and highlight the need for longitudinal and interventional studies. Keywords: adolescence, biological maturation, cardiovascular fitness, clustering analysis, explainable machine learning, functional capacity, peak height velocity, physical fitness

What kind of evidence is this?

Supporting AI learning science · Survey / observational study

Stage: Secondary school. Screening labels describe reported context and study-design language. They are not a risk-of-bias rating or proof of effectiveness for an age band.

Participant ages · source passage

Methods In this cross-sectional school-based study, 3,166 Tunisian adolescents (11–18 years) were categorized as pre-PHV (n = 959), around-PHV (n = 953), or post-PHV (n = 1,254).

Study methods · source passage

This study was conducted in accordance with the Declaration of Helsinki. The research protocol was reviewed and approved by the Institutional Scientific and Ethical Committee of Physical Education and Sport at the University of Jendouba, Tunisia (reference: PHS-31/2022). The study procedures complied with the relevant requirements and procedures governing research in sports medicine and exercise science (39).

Limitations and uncertainty · source passage

Several limitations must be acknowledged. The cross-sectional design precludes causal inference and limits understanding of individual developmental trajectories. Longitudinal studies are needed to determine whether the relative hierarchy of predictive contributors and exploratory profile assignments change within individuals during puberty (63, 74). Functional capacity was assessed using a submaximal test; incorporating maximal effort measures could provide additional information (92). Biological maturation was estimated using anthropometric equations rather than direct hormonal assessment, which may have introduced misclassification (53, 93, 94). This anthropometric estimation may also introduce structural dependence with body-size predictors, warranting cautious interpretation of their predictive contributions. Physical activity was self-reported, and may therefore be subject to recall bias (48). In addition, the absence of systematic repeated anthropometric measurements may have introduced residual random measurement variability, although standardized procedures and direct supervision were used to reduce this risk. Finally, because all participants were school-based adolescents from Tunisia, the generalizability of the findings to other geographic, ethnic, socioeconomic, athletic, and clinical populations may be limited.

Design the system before the pilot

Use this study to inform a learning mechanism or implementation question. It is not direct evidence that autonomous agents improve learning. Keep the proposed agent behavior separate from what the source actually tested.

Use common rubrics, blind scoring where feasible and a fresh unaided reasoning task. Document permissible assistance and avoid treating AI-detector scores as proof of misconduct.

A 90-day district pilot

Days 1–15 · Appraise. Have the curriculum lead and a teacher read the methods and limitations. Record the study design, sample, comparator, outcomes and evidence gaps. Define a local question that the study can help inform.
Days 16–30 · Design. Ask learners to attempt a problem unaided, compare an agent hint, explain a revision and verify the final reasoning against a trusted source. Make the permitted use explicit for each assignment. Name an accountable owner and set the baseline, success threshold and budget before implementation.
Days 31–60 · Test. Start with a small voluntary cohort, preserve an accessible non-AI route and require teacher review. Use synthetic or de-identified examples; approve data handling and tool terms through district processes.
Days 61–90 · Decide. Measure unaided transfer and delayed retention, calibration of confidence and the ability to identify an incorrect AI explanation. Record whether help-seeking becomes answer copying. Compare with baseline, document uncertainty and stop or revise if learning worsens, access is unequal or critical errors persist.

The board decision brief

Report the learning problem, evidence quality, participating population, total cost including review time, measured benefit, errors, access differences and remaining uncertainties. Make a dated recommendation to stop, revise or expand. A positive perception survey alone does not establish learning improvement.

Attribution and reuse

Hatem Ghouili; Amel Dridi; Nejmeddine Ouerghi; Jaouher Hamaidi; Wissem Dhahbi; Halil İbrahim Ceylan; Valentina Stefanica; Raul Ioan Muntean; Ismail Dergaa; Anissa Bouassida; Noomen Guelmami. School-based predictive contributors to six-minute walk test performance across pubertal stages: an explainable machine learning study in Tunisian adolescents. Frontiers in Sports and Active Living (2026-09-18). https://doi.org/10.3389/fspor.2026.1809681

Companion prepared 2026-10-30. Original authorship and DOI are preserved. The abstract is an attributed source extract; the implementation plan is new AI-assisted editorial material, offered under the same CC BY 4.0 license. This guide has not undergone independent peer review.