An Intelligent Air Monitoring System For Pollution Prediction: A Predictive Healthcare Perspective

Abstract The extensive potential of Internet of Things (IoT) technology has enabled the widespread real-time perception and analysis of health conditions. Furthermore, the integration of IoT in the healthcare industry has resulted in the development of intelligent applications, including smartphone-based healthcare, wellness-aware recommendations and smart medical systems. Building upon these technological advancements, this research puts forth an enhanced framework designed for the real-time monitoring, detection and prediction of health vulnerabilities arising from air pollution. Specifically, a four-layered model is presented to categorize health-impacting particles associated with air pollution into distinct classes based on probabilistic parameters of Health Adversity (HA). Subsequently, the HA parameters are extracted and temporally analyzed using FogBus, a fog computing platform, to identify vulnerabilities in individual health. To facilitate accurate prediction, an assessment of the Air Impact on Health is conducted using a Differential Evolution-Recurrent Neural Network. Moreover, the temporal analysis of health vulnerability employs the Self-Organized Mapping technique for visualization. The proposed model’s validity is evaluated using a challenging dataset comprising nearly 60 212 data instances obtained from the online University of California, Irvine repository. Performance enhancement is assessed by comparing the proposed model with state-of-the-art decision-making techniques, considering statistical parameters such as temporal effectiveness, coefficient of determination, accuracy, specificity, sensitivity, reliability and stability.