Effects of Different Stiffness Settings on Isokinetic Quadriceps Training in Children With Spinal Muscular Atrophy: A Randomized Within-Participant, Contralateral-Limb-Controlled Trial

Spinal muscular atrophy is characterized by progressive muscle weakness and impaired motor function. Quadriceps weakness limits knee extension, transfers, standing, and other activities of daily living. A portable isokinetic knee training robot has previously been shown to provide safe, quantifiable resistance training in children with spinal muscular atrophy. The present study will investigate whether the stiffness of the human-robot interaction affects rehabilitation outcomes.

Participants will be randomized to one of two laterality schemes: (A) compliant setting for the left leg and rigid setting for the right leg, or (B) rigid setting for the left leg and compliant setting for the right leg. An independent statistician will generate the random sequence, which will be concealed in sequentially numbered, opaque, sealed envelopes. The study is open label.

Before training, participants will undergo bilateral quadriceps magnetic resonance imaging, femoral nerve conduction testing, biomechanical assessment, and determination of the individually optimized compliant stiffness. Under the compliant setting, stiffness will be selected within the predefined safe range as the level that produces the highest mean peak knee-extension torque across five repetitions while the movement remains painless, without obvious fatigue, and with a stable trajectory.

Training will continue for up to 8 weeks, with no more than five sessions per week. During each session, the robot will be worn on one leg at a time while the contralateral leg is stabilized. Participants will perform seated knee extension and flexion at a target angular velocity of 45 degrees per second, adjusted when clinically necessary. The initial knee angle will be 90 degrees, and movement will occur within each participant's maximum voluntary pain-free range of motion. A standard set will consist of 10 continuous extension-flexion repetitions, with at least six sets per leg and approximately 1 minute of rest between sets. The number of repetitions, sets, and rest duration may be adjusted according to participant tolerance, but the training dose will be matched between the two legs. Total session duration will be approximately 35 minutes.

The robot software will record training data, including knee-extension peak torque and joint angle. After the intervention, bilateral quadriceps magnetic resonance imaging, femoral nerve conduction testing, and biomechanical assessments will be repeated. Participants will be evaluated in person before and after the intervention and contacted weekly by telephone during training to assess knee pain, discomfort, fatigue, and subjective changes in strength.