In Mat Pilates training, exercise balls and elastic bands will be used. The exercises will be done with 10 repetitions for the first 4 weeks, and with 20 repetitions for the last 4 weeks. The difficulty level of the exercises will be increased in the following weeks according to the patient's condition.

In the Reformer Pilates group, the difficulty level of the exercises will be provided by different positions, and the springs' resistance will be increased.

The control group will perform the relaxing exercise at home.

In Mat Pilates training, exercise balls and elastic bands will be used. The exercises will be done with 10 repetitions for the first 4 weeks, and with 20 repetitions for the last 4 weeks. The difficulty level of the exercises will be increased in the following weeks according to the patient's condition.

Data will be collected using an EMG-IMU system (Trigno Avanti, Delsys), with an accelerometer detecting perturbation onset and EMG recording tibialis anterior activity (1000 Hz, 16-bit; SENIAM placement).

Participants will stand barefoot on a force platform (feet shoulder-width apart) while perturbations are applied via a 4-kg pendulum. They will receive impacts with both hands at shoulder level and maintain balance. Three trials will be performed.

EMG signals will be rectified, low-pass filtered (100 Hz), and analyzed from -600 to +500 ms around perturbation onset (T0). APAs (-100 to +50 ms) will be baseline-corrected and normalized, with mean values used for analysis.

CoP data will be recorded during 30-s trials with perturbation at 15 s. CoP ellipse area will be calculated, and the mean of three trials will be analyzed.

Postural stability will be assessed using a force plate (K-Force, K-Invent), a validated and reliable tool for CoP analysis. Measurements will be taken under eyes-open and eyes-closed conditions, each lasting 30 seconds and repeated three times, with 10-second rests between trials and a 1-minute break between conditions.

The primary outcome will be CoP ellipse area (mm²), representing postural sway. Mean values from the three trials for each condition will be used for analysis.

Spatiotemporal gait parameters will be assessed using a wearable inertial sensor system (G-Walk, BTS Bioengineering, Italy). Participants will walk a 10-m distance at a self-selected comfortable speed in three trials. Acceleration data will be sampled at 100 Hz and transmitted via Bluetooth to G-Studio software for analysis.

Measured parameters will include gait velocity, cadence, stride length, gait cycle duration, and double support time (% of gait cycle). The mean of the three trials will be used for statistical analysis.

Functional mobility will be assessed using the Timed Up and Go (TUG) test, a reliable and valid clinical measure. Participants will stand up from a chair, walk 3 m, turn, return, and sit down. After one familiarization trial, three trials will be recorded, and the mean value will be used for analysis.