In the light of the development of polydigital hand prosthesis, we recently explored an intuitive control mode by surface EMG associated to phantom limb movements (PLM) in upper arm amputees (Jarrassé et al., 2017a, b). As little was known about PLM, we described types and characteristics of upper limb PLM and explored some factors potentially influencing these in a population of upper limb amputees (De Graaf et al., 2016; Touillet et al., submitted). We showed, among others, that one persistent characteristic of PLM is the associated fatigue, which potentially is a problem when using PLM for prosthesis control since fatigue diminish the mobilization capacity as well as the stability of the EMG signals and thus interferes with the prosthesis control.

In the present study, we explored whether PLM training (not to confound with learning) can increase mobilization capacity and stabilize EMG signals. Five trans-humeral amputees volunteered for training of all their types of PLM at home on a daily basis during about 6 weeks. Kinematics (via the intact limb imitating the phantom movements) as well as EMG signals from the residual muscles were recorded just before and right after the training period.

None of the participants experienced (phantom or residual limb) pain related to the training. Three participants increased the number of different PLM they could execute. All of them increased the amplitude and velocity, and thus diminished the cycle duration, as well as the number of cycles they could execute before the PLM blocked by fatigue. The amplitude of the EMG increased and became more specific to the type of PLM.

In conclusion, as is the case for intact limbs, training of PLM improves motor performance and stabilizes the EMG signals associated to PLM execution. This is encouraging for future PLM-based prosthesis control.