Strange interference in the EMG sensors of bionic prostheses. Who faced?

Question

Noise suppression

A strange situation - the interference of the EMG sensor of domestic production. We have 6 pieces of electromyographic sensors. We launch on our own boards. On some “good” employees, the expected signals are caught, on some “noisy” ones, only noise. "Good" employees are about 10%, "noisy" 90%. If the “good employee” is touched by the “noisy” one, the normal signal becomes noise for the duration of the contact. Rotating the trimmer doesn't help.
Who's to say why? That fire and pizza from us).