Soft Computing Framework for State-Level Prediction of Biomedical Waste Generation in India: An ANN-Based Multi-Factor Analysis

Abstract This paper explores the application of an artificial neural network (ANN) to forecast annual biomedical waste (BMW) production at the state level in India. Traditional methods are inadequate for precise prediction due to the varied sources of BMW, and advanced techniques are required to overcome such. The study employed a feed-forward neural network with sigmoid hidden and output neurons to create a model for the estimation. At the same time, Levenberg-Marquardt backpropagation was used to optimize the network. Various health, social demographic, and economic factors were used as input variables, and ten models were developed using sensitivity analysis. An annual dataset covering the years 2010–2020 was used to train the algorithms. This scenario marks the first application of BMW forecasting using whole Indian statistics, providing a notable advancement in the field. The statistics showed a significant increase in BMW generation, from 410.33 tons/day in 2011 to 628.3 tons/day in 2020, highlighting the urgency of predictive models in consideration of the growing population and waste yield. Key variables analyzed include hospital beds, state gross domestic product trends, healthcare facilities (HCFs), urban population, authorized HCFs, the number of operational common BMW treatment facilities, and HCF authorization applications. Models' performance was evaluated using mean square error (MSE) and determination coefficient (R), with model seven exhibiting the best performance with an MSE of 0.0032 and R of 0.975. The study concludes that the ANN approach can effectively predict future BMW production in India and other cities facing similar issues for adequate planning and improvement approaches in the healthcare system and waste management. The findings emphasized the importance of advanced modeling approaches for BMW generation while providing significant and applied implications for policy and environmental management concerns.