Abstract
Abnormal gait and postural instability are common disorders in people affected by Parkinson's disease (PD). This paper proposes an embedded cyber-physical system for the identification and the real-time extraction of highly selective diagnostic indexes for PD patients. A noninvasive wearable and wireless architecture for both gait analysis and postural instability detection has been proposed and implemented on a programmable hardware. The combined analysis of electroencephalography and electromyography allows studying the motor cortex activity through the movement-related potentials, determining a novel set of indexes that could be used for the PD diagnosis and classification. In a future perspective of an application-specific integrated circuit implementation, the real-time data processing has been fully realized on the Altera Cyclone V field-programmable gate array (FPGA), without interactions with embedded processor architecture. Referring to an Altera Cyclone V SE 5CSEMA5F31C6N device, the whole implemented architecture exploits 90% of the available FPGA adaptive logic modules, 74% of the manageable registers, and 10.3% of the total memory, as well as 29.7% wires utilization. Furthermore, the system is able to provide the outputs in about 57 ms with a dynamically power dissipation of 89 mW. The platform has been tested in vivo on two Parkinson's patients and two healthy subjects (control group) covering three typical diagnostic scenarios: PD versus controls, drug treatment evaluation, and involuntary movements detection. © 2018 IEEE.