Hardware Components in Intelligent Systems for Parkinson’s Disease Diagnosis

Abstract

This paper proposes an IT-based diagnostic system for Parkinson’s disease, supported by the OSTIS platform. The system combines a domain ontology and knowledge base to formally represent and manage medical information related to Parkinson’s disease. A specialized hardware-accelerated processing module based on FPGA is integrated to enable real-time analysis of patient data. Discrete wavelet transform and a gated recurrent unit (GRU) neural network are used for extracting and analyzing motor features. The hardware implementation was realized using Xilinx Vitis HLS, allowing efficient deployment of wavelet and GRU modules. Preliminary experiments on open datasets show that the system achieves approximately 95% accuracy in recognizing speech and motor patterns associated with Parkinson’s disease.As a result of using FPGA, processing is accelerated by 20 times compared to the CPU implementation.