Artificial Intelligence-Enabled Monitoring Framework for Aquatic Ecosystem Changes Prediction and Conservation Decision Support

Aquatic ecosystems face increasing levels of pollution, environmental variability, and ecological stress, making intelligent monitoring and future conservation planning necessary. ThisThis research seeksseeks to design an AI-based framework to predict changes in aquatic ecosystems, evaluate risks in these environments, and support conservation decisions. Our framework combines multi-source aquatic environmental observations together with data preprocessing, feature engineering, AI-based prediction, Ecosystem Risk Index (ERI) evaluation, and decision-making support for conservations. Random Forest, XGBoost, and Long Short-Term Memory (LSTM) models were compared with the proposed framework on Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the coefficient of determination (). The proposed framework outperforms the other methods in terms of MAE (0.184 vs. 0.286 in Random Forest; 0.231 in XGBoost; 0.247 in LSTM) and () (0.94 vs. 0.86 in Random Forest; 0.91 in XGBoost; 0.89 in LSTM). Environmental Risk Index (ERI) values ranged between 0.24 and 0.71, with the maximum value obtained for the period P8, showing relatively high environmental risk. The findings reveal the capability of incorporating an AI-based forecast with ecosystem risk assessment to detect changes in aquatic environment conditions and guide the process of monitoring efforts. Future research is needed to validate the concept via field observations, IoT technology, remote sensing, and other data sources.