,
Research Scholar, Department of Computer Applications, B. S. Abdur Rahman Crescent Institute of Science and Technology , Chennai, Tamil Nadu , India
Assistant Professor (Senior Grade), Department of Computer Applications, B.S.Abdur Rahman Crescent Institute of Science and Technology , Chennai,Tamil Nadu , India
Heterogeneous, high-dimensional, and complex clinical data have remained a key challenge for early sepsis detection in intensive care units. This paper introduces a Hybrid GAP-HMSOA-GNN -COMPOSER architecture that incorporates a graph-based relational model and metaheuristic optimization to learn and generate accurate, reliable sepsis predictions. The Graph Aggregation Process (GAP) allows the model to learn temporal and relationship relationships among patient features, and the Human Mental Search Optimization Algorithm (HMSOA) is used to systematically optimize model parameters and hyperparameters to promote generalization and eliminate local minima. The given framework was tested on a clinical electronic health record dataset using several common statistical measures, including Accuracy, Precision, Recall, F1-score, Area Under the ROC Curve (AUC), and Matthews Correlation Coefficient (MCC). The model achieved Accuracy = 94.2, Precision = 92.8, Recall = 93.5, F1-score = 93.1, AUC = 0.96, and MCC = 0.88, outperforming baseline classifiers such as Random Forest, SVM, XGBoost, LSTM, and standard COMPOSER. Ablation experiments verified the role of each part: deletion of GAP or HMSOA caused a statistically significant performance degradation (AUC decrease of 3-5%), indicating a synergistic effect of aggregating relational features and metaheuristic optimization. The strength of the proposed model was confirmed through comparative analysis across heterogeneous patient groups and demonstrated its practical applicability. These findings reveal that the hybrid GAP-HMSOA-GNN-COMPOSER model is a statistically sound, interpretable, and generalizable approach for early sepsis detection, enabling timely intervention and better patient outcomes.
This is an open access article distributed under the Creative Commons Attribution Non-Commercial License (CC BY-NC) License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The statements, opinions and data contained in the journal are solely those of the individual authors and contributors and not of the publisher and the editor(s). We stay neutral with regard to jurisdictional claims in published maps and institutional affiliations.
