Lists
Lists
Komala, C. R., Virdee, Bal Singh, Khanna, Ashish, Vidhya, R. G. and Sivashankar, S. (2026) Adaptive and low-latency COVID-19 monitoring with PO-ADMGNN in fog-assisted healthcare system. In: 2025 International Conference on NexGen Networks and Cybernetics (IC2NC), 1-3 December 2025, Erode, India.
Cardiovascular disease continues to be one of the leading causes of death globally, and forecasting its occurrence in diabetic patients is extremely difficult with the nature of medical, behavior, and heterogenous health data. The traditional ML and shallow DL models are inadequate with noisy, unbalanced, and multi-source data, decreasing their precision and generalizability. To address these issues, the current study suggests a new framework called Pufferfish Optimization - Adaptive Dynamic Multi-Graph Neural Network (PO-ADMGNN) that has been specially built to yield robust and precise heart disease prediction. The methodology begins with detailed preprocessing of two popular datasets and then using the Golden Jackal Optimization Algorithm (GJOA) for feature selection to identify best informative features. The central ADMGNN model includes weighted multi-source data fusion, attention-based convolution-pooling layers for detailed feature extraction, and a temporal module for spatial - temporal pattern capturing. An ensemble stacking technique consists of several base learners, and a Support Vector Machine is the meta-learner. Hyperparameters are optimized using the Pufferfish Optimization Algorithm to achieve maximum predictive performance. Experimental tests validate the performance of the proposed system. On Dataset 1, the model scored 98.0% accuracy, 97.0 precision, 99.0% recall, and a 95.0% F1-score. On the Cleveland dataset, it scored 99.0% accuracy, 98.0% precision, 97.0% recall, and 98.0% F1-score. With hyperparameters set (learning rate=0.001, batch size=32, dropout rate=0.2), the model had rapid convergence—training for 120 seconds and testing for 8 seconds over Dataset 1, and 65 seconds training and 5 seconds testing over Cleveland.
 |
View Item |