Reformer pilates improves posture and functional movement: a randomised comparison of AI-assisted and traditional personalisation
Year: 2026
Type: article
Study designs: Randomized Controlled Trial, Clinical Trial
Abstract: Postural deficits and functional movement limitations are growing public health concerns. While Reformer Pilates is known to improve alignment and movement quality, the efficacy of AI-driven weekly personalisation remains unexplored. This randomised pre-test/post-test study compared an 8-week ChatGPT-assisted Reformer Pilates programme, adapted weekly, against a fixed traditional protocol of equivalent volume. Thirty healthy adults (20–45 years) with prior Pilates experience were randomly assigned to an AI-assisted group (AIG, n = 15) or a traditional group (TG, n = 15). Participants trained twice weekly for 60 min over 8 weeks. The AIG followed ChatGPT-generated programmes revised weekly based on individual performance metrics; the TG followed a fixed instructor-designed protocol. Outcomes were assessed via the Advanced Posture Evaluation and Correction System (APECS) and the Functional Movement Screen (FMS). Data were analysed using a 2 × 2 mixed ANOVA with Holm–Bonferroni correction, with confidence intervals for all effect sizes and a baseline-adjusted ANCOVA as a sensitivity analysis. Significant time main effects were observed for all five posture parameters (Shoulders: F = 28.47, η² p = 0.50; ASIS: F = 14.14, η² p = 0.34; Feet: F = 9.66, η² p = 0.26; Knees: F = 9.16, η² p = 0.25; Ears: F = 6.13, η² p = 0.18; all p_adj ≤ 0.020) and all seven FMS subtests (η²p range 0.37–0.49; p_adj < 0.001). Neither group main effects nor Time × Group interactions reached significance (all p_adj ≥ 0.48), and no baseline-adjusted between-group difference survived correction, so improvements were comparable across both interventions. An 8-week Reformer Pilates programme produced large, clinically meaningful gains in posture and functional movement in both groups. ChatGPT-based weekly adaptation offered no statistically significant advantage over a fixed instructor-designed protocol under the conditions of this trial. Future research should involve larger samples, longer durations, and sensor-derived biometrics to further evaluate the potential of AI in exercise prescription. Chinese Clinical Trial Registry ChiCTR2600126057. Registered 03 June 2026. Retrospectively registered.
Authors: Bilge Gaye Yılmaz, Naci Kalkan
Institution: Manisa Celal Bayar University
Language: English
Field: Health Professions
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,890
OpenAlex ID: W7153340124