Personalised healthcare model for monitoring and prediction of airpollution: machine learning approach
Veerawali Behal, Ramandeep Singh
Abstract
The drastic increase in atmospheric pollutants has resulted in the prevalence of hazardous diseases like Asthma, Ischaemic heart disease, and Pulmonary disease around the world. IoT technology has the capability to acquire and monitor air quality parameters in the ambient environment of an individual. Inspired from these aspects, this paper proposes an IoT-based automated framework for monitoring and predicting air quality parameters like benzene using machine learning technology. Specifically, this study incorporates an Adaptive Neuro-Fuzzy Inference System (ANFIS) technique to predict the air quality in the form of Level of Pollutant (LoP) and modified Air Quality Index (m-AQI). Moreover, for the optimisation of ANFIS technique, Differential Evolution (DE)-inspired algorithm has been proposed for overall enhancement of system accuracy. In order to validate the proposed model, numerous experimental simulations were performed over four challenging datasets and results were compared with several state-of-the-art models. Comparative analysis shows improved statistical values for the presented model in terms of Accuracy (97.3%), Coefficient of Determination (94.01%), and Root Mean Square Error (2.4%). In addition to these, enhanced values of performance estimators like Reliability (92.36%) and Stability (76.00%) were estimated for depicting overall efficiency of the proposed system.