Group-by-time interactions favored the intervention group over the control group on gait, balance and psychological outcomes