Scalable predictive framework for environmental pathogen control in land-based aquaculture

Land-based aquaculture requires scalable treatment systems capable of anticipating and mitigating pathogenic risks under changing environmental conditions. In this study, we collected meteorological and bacterial data and performed correlation analyses to identify key relationships, which guided the development of an integrated, predictive treatment system. This system combines a modular total suspended solids–pathogen removal system (TSS–PRS), composed of sediment filtration, UV disinfection, and oxygen dissolution, with a deep learning-based multi-layer perceptron (MLP) model to improve water quality and forecast pathogen dynamics. The TSS–PRS effectively reduced TAN (41.1%), bacterial activity (BQV, 74.5%), and turbidity (72.8%). It also successfully eliminated hazardous fish pathogens, including Photobacterium damselae, Tenacibaculum maritimum, Vibrio harveyi, and Enteromyxum leei. The MLP model further indicated that bacterial activity markedly increased under optimal conditions of turbidity (100 NTU), pH (7.97), and water temperature (27.5 °C).