Optimized Triple Memristor Hopfield Neural Network fostered Automated Outbreak Prediction of Epidemic Diseases using Internet of Things

Abstract

The paper presents a comprehensive system for predicting and preventing epidemic outbreaks using the Internet of Things (IoT), combined with an Optimized Triple Memristor Hopfield Neural Network (TMHNN) and the Northern Goshawk Optimization (NGOA) algorithm. This approach, referred to as AOPE-TMHNN-NGOA-IoT, gathers real-time health data from individuals through wearable IoT devices. Critical health parameters like body temperature, blood pressure, and heart rate are monitored using a Dengue dataset. The collected data is first subjected to pre-processing using Nanoplasmonic Ultra Wideband Bandpass Filtering (NUWBF) to remove noise and outliers. Following this, feature extraction is performed using the Two-sided Offset Quaternion Linear Canonical Transform (TOQLCT), which extracts the essential features for disease detection. These features are then fed into the TMHNN, which is optimized by the NGOA algorithm to classify individuals as either infected or healthy. When an infection is detected, the system immediately alerts the individual via the IoT device, enabling early intervention. The AOPE-TMHNN-NGOA-IoT system is implemented using Python and demonstrates superior performance compared to existing methods, including AOPE-SVM-IoT, AOPE-KNN-IoT, and AOPE-NB-IoT. Specifically, the proposed system achieves 25.45%, 19.12%, and 27.11% higher accuracy, and 15.36%, 21.55%, and 18.74% higher specificity than these techniques, respectively. By integrating IoT, advanced neural networks, and optimization algorithms, AOPE-TMHNN-NGOA-IoT provides a robust, efficient solution for predicting and preventing epidemic outbreaks. This system significantly enhances health monitoring, offering a proactive method for controlling the spread of infectious diseases globally.