COVID-19 Patient Secure Classification Based on ANN and Risk Analysis

The introduction of the coronavirus (COVID-19) has resulted in higher surveillance costs, international hospital overcrowding, and disastrous diseases for the public. In order to treat and diagnose patients, wireless body networks (WBAN) have become essential. In computer-assisted rehabilitation and health surveillance, WBAN offers a wide range of applications, including those pertaining to emergency medical response systems, such as in the present pandemic situation. The patient's physiological characteristics are measured by biomedical sensor nodes, and the results are stored on a system called a local therapy unit. Only authorized users (doctors and other healthcare professionals) can remotely view the data from these patients as long as they are linked to their smart devices. However, in extreme circumstances, abnormal or incorrect measurements caused by damaged or malicious sensor nodes may result in incorrect diagnosis and even patient death. Patient privacy is in danger when these new technologies are used without taking data security into account. Thus, it is crucial for healthcare apps to retain anonymity. In order to resolve these problems, we suggest an effective healthcare system based on distributed agents. This system will ensure the secure collection and transfer of COVID-19 patient data to healthcare experts and will make it easier to identify and diagnose fatal irregularities. Our main contribution is to provide distributed real-time processing, learning, and risk analysis capabilities in order to deliver an intelligent and secure system. Agents are in charge of locating legitimate patient datasets from Tunisian doctors. In order to do this, we calculated risks related to sensors, signals, and the wireless medical network, sent data via Telnet and NetCat, employed ANN as the learning method, and securely sent data using the AES encryption protocol security. Our method exhibited performance values of 99% accuracy, 100% recall, and an F-score of 99.4% with a running time of 1.50 seconds.