• machine learning
• safety
• control
• neuroprosthesis
• peripheral nerve stimulation
• hardware
• setup
• biomechanics

My thesis belongs to the framework of a pilot study aimed at restoring hand grasping in two tetraplegic patients, by means of intraneural peripheral stimulation. The objective of my work is to develop a system to evaluate efficacy and safety of the neuroprosthesis. It articulates in two phases. The development of a software suite to characterise the motor response to the stimulation, and implementing a software for the real-time voluntary control of the neuroprosthesis. I performed a safety test of the experimental setup, to inspect the possible causes of software malfunction. Finally, I collected a database from 5 healthy volunteers performing grasping tasks, that will be useful to compare with the data acquired from the patients in the clinical study. The GUI I developed allows to set the electrical stimulation and to simultaneously record EMG, grip force and hand kinematics. The software for voluntary control acquires data from surface EMG and a pillow switch to modulate the electrical stimulation in real-time. I conducted the risk analysis by simulating many cases of software malfunction and provided measures to keep control over the ongoing stimulation and ensure it stays inside a safe range. The database comprises hand kinematics and surface EMG from the forearm, and I validated it through a principal component analysis to assess whether different grasps can be separated thanks to the available features.