Reformer pilates improves posture and functional movement: a randomised comparison of AI-assisted and traditional personalisation

This study compared the effects of a traditional fixed Reformer Pilates programme and a ChatGPT-assisted, weekly-adapted Reformer Pilates programme on postural alignment and functional movement quality in healthy adults. Thirty volunteers aged 20–45 years with a minimum of 6 months of prior Pilates experience were randomly allocated to an AI-assisted group (n=15) or a traditional group (n=15). Both groups trained on the Combo Cadillac Reformer twice weekly for 60 minutes over 8 consecutive weeks. The AI-assisted group received a programme revised each week by ChatGPT based on individual flexibility, balance, and muscular endurance data; the traditional group followed a fixed instructor-designed protocol throughout. Postural alignment was assessed using the Advanced Posture Evaluation and Correction System (APECS) at five anatomical landmarks (ears, shoulders, ASIS, knees, feet), and functional movement quality was assessed using all seven subtests of the Functional Movement Screen (FMS). The primary hypothesis was that AI-assisted weekly personalisation would produce superior improvements compared with the traditional protocol. Data were analysed using 2×2 mixed ANOVA with Holm-Bonferroni correction. Both groups showed statistically significant and large-magnitude improvements in all posture parameters and all FMS subtests over the 8-week period. However, neither the group main effect nor the Time × Group interaction reached significance for any outcome, indicating that the two programmes produced equivalent improvement trajectories. These findings suggest that the core principles of Reformer Pilates are the primary driver of adaptation, and that ChatGPT-based weekly programme personalisation did not confer a statistically significant advantage over traditional instruction within this sample size and intervention duration. This study is derived from the master's thesis of Bilge Gaye Yılmaz (Manisa Celal Bayar University, 2025; supervisor: Assoc. Prof. Dr. Naci Kalkan).