Abstract

Shrimp farms play an important role in the international seafood value chain, but due to disease outbreaks, these farms can be easily affected, causing disastrous economic and ecological consequences. This study aims to apply Long Short-Term Memory (LSTM) deep learning models to forecast disease outbreaks in shrimp farming systems. Since there is historical data on environmental conditions, water quality, and recorded disease values, the LSTM model was used to learn time-series and trend patterns to prevent future outbreak risks. The main results are mostly grounded in accuracy, precision, and other metric indicators, and it is apparent that they perform better than the traditional methods. The proposed model is optimistic and shifts the infection control approach from reactive to proactive, using predictive control signals to prevent intervention once the disease has spread and reduce economic costs. This research demonstrates how modern artificial intelligence can be implemented in aquaculture, and further questioning of policy frameworks will enable the optimization of intelligent, environmentally friendly, and resilient agricultural models in cultivable aquaculture to achieve sustainability. By implementing bioinformatic pathosystems, a more direct multivariate analysis strategy could be established, creating a reusable, customizable template for aquaculturists interested in enhancing shrimp health, boosting agricultural output, and advancing biosecurity. The new research will connect sensor data to the model and will add more diseases and regions to the predictive analytics.