Improved prosthetic hand control with concurrent use of myoelectric and inertial measurements.

The real-time group was small and mostly able-bodied, with only one amputee participant, so the amputee result cannot be treated as evidence for a population. Able-bodied participants had their fingers bandaged closed to mimic limb loss, which may not reproduce the muscle signals, socket fit or daily use of an actual prosthetic user. Training and testing used the same arm postures for the same grips, which may have favoured decoders that use posture-dependent inertial signals. The study tested a short laboratory task and reported technical control metrics, not patient-reported function, independence, safety or long-term outcomes. One amputee's offline data collection was interrupted early because of a power supply failure, so that participant did not perform the final two movements.

Funding was from Engineering and Physical Sciences Research Council, Biotechnology and Biological Sciences Research Council and Microsoft Research.

Do not conclude that adding inertial sensors makes prosthetic hands clinically better for amputees. The real-time evidence is mainly from able-bodied users whose fingers were bandaged, and the amputee finding comes from a single participant in a short laboratory task.