Background <p>Feedback-based gait training has been shown to improve gait performance. Feedback gait training using inertial measurement units (IMUs) has been implemented in clinical settings but is limited to examining immediate effects; long-term effects on gait parameters such as gait speed remain unclear. This pilot study aimed to investigate the short-term effects of real-time feedback gait training using IMUs on gait parameters in older inpatients.</p> Methods <p>Seven older inpatients participated in this within-subject study with baseline, control, and intervention periods. During the intervention period (five times per week over one week), participants received real-time auditory feedback using IMUs to increase ankle plantar-flexion angle or leg-extension angle, defined as the ankle position relative to the hip in the sagittal plane during the late stance. The tilt angles and accelerations of the pelvis and lower-limb segments were measured using seven IMUs at each time-point. A one-factor repeated-measures analysis of variance, Friedman test, and post hoc tests were conducted to examine the short-term effects of gait training. One-dimensional statistical parametric mapping (SPM1d) was used to assess phase-specific changes across the gait cycle.</p> Results <p>Gait speed and stride length increased during the feedback period with large effect sizes (<i>p</i> = 0.047; <i>r</i> = 0.91). Ankle plantar-flexion angle on the feedback side differed across baseline, pre-training, and post-training assessments (F = 4.46, <i>p</i> = 0.036; η² = 0.43), but post hoc tests showed no differences between time-points (<i>p</i> ≥ 0.065). The leg-extension angle did not differ between time-points on the feedback side (F = 1.40, <i>p</i> = 0.283; η² = 0.19). SPM1d analysis of gait kinematics, including ankle plantar-flexion and leg-extension angles, identified significant clusters during the late stance phase.</p> Conclusions <p>Feedback gait training using IMUs demonstrated potential for increasing gait speed, with substantial effect sizes and phase-specific kinematic changes identified using SPM1d. These preliminary findings suggest that feedback-based gait training could modify both spatiotemporal and kinematic parameters, and provide initial evidence regarding the feasibility and potential for improving gait ability in older inpatients in clinical settings.</p> Trial registration <p>University Hospital Medical Information Network Clinical Trial Registry (UMIN-CTR 000051328).</p>