Publication

Abstract : The improper management of biomedical waste in healthcare organizations represents escalating environmental, occupational, and operational challenges. More than 5.9 million tonnes of medical waste are generated annually all over the world, and approximately 15 percent of this waste is classified as hazardous. This poses serious health hazards, infection risks, and high treatment costs. The study proposes a machine-learning based cyber-physical framework that enables smart hospital waste segregation, elevates worker safety, and enhances sustainable healthcare operations with resource efficiency. A two-stage deep learning pipeline integration of YOLOv8, EfficientNet-B0, and ResNet-50 accurately identifies hazardous, non-hazardous, biodegradable and non-biodegradable waste streams. The architecture is an edge AI system that is tested and validated at the software stage and is driven by the need to provide energy efficient inference and real-time decision making, hence making it applicable to the industrial deployment. The outcomes highlight great potential to decrease human exposure to infectious materials, decrease landfill and incineration loads and optimize waste lifecycle performance and lead to circular, resource-efficient, and environmentally sustainable healthcare waste management systems.