Cognitively Enhanced Tai Chi for Executive Function in Children With Autism Spectrum Disorder

Study design and allocation This single-center, three-arm parallel trial uses individual randomization stratified by original administrative classroom. The target sample was 165 children, with planned arm sizes of 66, 66 and 33. Recruitment closed with 158 eligible children enrolled. The final allocation retained a target ratio of 2:2:1 using prespecified integer quotas based on the locked enrollment list. Original classrooms are randomization strata; stable 4-6-child training groups are formed within intervention arms and may combine children from different original classrooms. The primary comparison is cognitively enhanced Tai Chi versus standard Tai Chi at month 6. Enhanced Tai Chi versus general exercise is supplementary; standard Tai Chi versus general exercise is exploratory.

Intervention and follow-up All three courses comprise 24 teaching weeks, 3 sessions per week and 45 minutes per session, totaling 72 sessions scheduled within 6 calendar months, with school-calendar and make-up buffers. Each session includes arrival and visual schedule (4 minutes), warm-up (5), three main activity blocks (10, 10 and 8), two scheduled rest breaks (2 minutes each), and cool-down and handover (4). A trained lead instructor and an ASD support teacher supervise each group, with additional support where required. Visual schedules, brief instructions, demonstrations, communication supports, opportunities to rest and participation-based feedback are shared across arms. Children may use seated adaptations. Necessary usual education, rehabilitation and medical care continue throughout. The courses ran from December 25, 2025 to June 25, 2026. Three months of follow-up followed the intervention, with visits at 1 month (month 7) and 3 months (month 9); brief contacts recorded independent practice, other exercise or rehabilitation, medication changes and safety events.

The enhanced Tai Chi course uses the same adapted six-movement library as standard Tai Chi, with added predictable start/pause cues, short action-sequence retention and preannounced neutral-symbol rule changes. Cognitive difficulty is adjusted individually, with at most two action cards, a short delay up to 5 seconds, or one rule change; demands are not combined into multiple simultaneous rules. The standard Tai Chi course uses visible demonstrations and a fixed visible action-sequence strip, without the added delayed-recall or arbitrary rule-switching tasks. The general exercise course uses predictable walking or stepping, gentle sandbag placement or throwing, ball rolling, and adapted basic strengthening, with fixed order and visible demonstrations.

Measurements Existing short computerized tasks are used without increasing trial counts, testing duration, difficulty or adding a mixed-switching phase. Flanker accuracy is the proportion correct across all validly presented formal trials. In the 1-back task, accuracy uses formal response-required trials; the initial reference image in each block is excluded. DCCS accuracy uses post-rule-change formal trials in the existing color-then-shape task. Accuracy ranges from 0 to 1, or 0-100%; higher values indicate more accurate task performance. The median reaction time in milliseconds among valid correct trials in the corresponding trial set is supportive. Incorrect and timed-out responses to validly presented formal trials count as incorrect; technical errors are recorded separately. An unevaluable or interrupted task is missing, rather than scored zero.

Computerized tasks and the complete 24-item CHEXI parent and teacher forms were collected at baseline and months 3, 6, 7 and 9. CHEXI inhibition/regulation uses the mean of 11 items and working memory/planning uses the mean of 13 items; each ranges from 1 to 5, with lower scores indicating fewer difficulties and no reverse scoring. The official Simplified Chinese parent SDQ with its impact supplement was completed at baseline and month 6, retaining its six-month recall period; parent and teacher questionnaires administered at the month 7 and month 9 visits that are not prespecified secondary outcomes are summarized descriptively. The SDQ impact score sums child distress and interference with home life, friendships, classroom learning and leisure; the range is 0-10 and lower is better. Duration and family burden are not included.

The eight secondary tests specified in the current analysis plan are the three month-6 task reaction times; month-6 teacher and parent CHEXI inhibition/regulation, month-6 teacher CHEXI working memory/planning, month-9 teacher CHEXI inhibition/regulation; and month-6 parent SDQ impact. DSM-5 diagnostic documentation, mainland Chinese WISC-IV full-scale IQ, the Chinese parent SRS, height and weight characterize eligibility and baseline status. They are not added as longitudinal efficacy outcomes.

Analysis The three month-6 accuracy comparisons between enhanced and standard Tai Chi form one Holm-adjusted primary family. The eight secondary tests form a separate Holm-adjusted family. Each accuracy outcome will be modeled using a binomial-logit mixed-effects model based on correct and incorrect counts, adjusting for baseline task accuracy, age and original classroom, with categorical follow-up time, group and group-by-time effects. Child and training-group random intercepts account for repeated measurements and group-based delivery. Marginal predicted accuracy, the enhanced-minus-standard difference in percentage points, 95% confidence intervals and adjusted P values will be reported. Log-transformed reaction-time medians and CHEXI scores use their respective linear mixed models; SDQ impact uses an ordinal-logit mixed model. Analyses retain participants according to original randomized allocation, with the missing-data sensitivity analyses specified in the current analysis plan. Actual short-task data and plausible treatment-induced group correlations will inform the planned power simulations; no completed numerical power result is claimed.

Registration history The initial registration materials were submitted to ClinicalTrials.gov on December 22, 2025, before the first guardian signed consent for enrollment on December 24, 2025. This record was updated in October 2026 to reflect the investigator-confirmed actual enrollment and study dates, course descriptions, and the revised outcome definitions and analysis plan.