IOT system architecture for the diagnosis of Parkinson's disease using neural networks and OSTIS

The purpose of this work is to develop an IT diagnostic system for Parkinson's disease (PD) with remote access based on the Internet of Things (IoT) network. Methods. The authors have developed a method for complex recognition of Parkinson's disease using machine learning, based on markers of voice analysis and changes in patient movements on known datasets. In the architecture of the Internet of Things network, a smartphone is the point of initial data collection and preprocessing, including extracting features from an audio recording of the patient's voice and his motor activity. Data is transmitted via a local Flask server, which acts as a channel for sending functional data to the Open Semantic Technology for Intelligent Systems (OSTIS) server. The OSTIS server processes the data received from the local Flask server and uses a neural network prediction agent to recognize BP. This agent downloads features and makes predictions based on a trained neural network, linking these predictions with knowledge in the OSTIS system, and stores them in a database. The result of the study is the architecture and algorithms of the IoT network. The workflow of the entire system includes data collection and preprocessing by the Internet of Things device, subsequent data transfer to the local Flask server, further forwarding to the OSTIS server, processing of the neural network model by a neural network predictor agent and, ultimately, linking the processed results to the knowledge graph and storing them in the system. The BP remote IT diagnostics system provides real-time processing of patient data, recognition of disease signs on the Internet of Things, support for advanced analysis and decision-making for further treatment.