AquaTwinCare: A Digital Twin-Inspired Framework for Aquatic Animal Healthcare

Water quality critically affects aquatic life, where even slight changes in chemical or physical conditions can cause stress or health decline. Real-time monitoring is challenging due to dynamic, distributed environments. This study introduces AquaTwinCare, a digital twin (DT)-based virtual replica of aquatic ecosystems for assessing and predicting animal health under pollution stress. The framework integrates environmental factors (dissolved oxygen, pH, temperature, turbidity, and pollutants) with biological indicators (respiration, locomotion, and stress responses) using hybrid convolutional neural networks (CNNs) enhanced with Bayesian inference. To ensure secure, reliable data, it employs a reputation-aware fault-tolerant consensus (RAFTC) protocol on a consortium blockchain. Validated on 91845 data instances, AquaTwinCare achieved reduced anomaly detection latency (9.80 s), high precision (84.25%), sensitivity (86.13%), specificity (86.19%), F-measure (85.15%), predictive fidelity (r² = 78%), low error (0.23%), and mean error (0.62). Overall, it enables proactive aquatic health management, early pollution-stress warnings, and scalable ecosystem governance.