Adaptive UV-C LED Dosage Prediction and Optimization Using Neural Networks Under Variable Environmental Conditions in Healthcare Settings

Abstract

Healthcare-associated infections affect approximately 1.7 million U.S. patients annually, resulting in over $30 billion in costs. This paper presents an adaptive neural network framework for real-time UV-C LED dosage optimization, addressing CDC priorities for hospital environmental safety. The system integrates multi-sensor monitoring with artificial neural networks to dynamically adjust irradiation based on temperature, humidity, distance, and surface properties. Experimental validation demonstrates a 36.2% energy reduction while maintaining 4.09 ± 0.3 log pathogen inactivation, supporting the DOE’s sustainable healthcare initiatives. The algorithm achieves an R² of 0.943, a MAPE of 6.2%, and a latency of 8.6ms, making it suitable for FDA-regulated medical devices. These results indicate that the proposed adaptive UV-C control framework can maintain a≥4-log₁₀ pathogen reduction while reducing redundant exposure time and energy consumption, supporting hospital infection control and energy efficiency priorities in U.S. healthcare environments, and aligning with emerging expectations for transparent, auditable AI-enabled disinfection systems.