Fusion-based CNN-SVM Hybrid Model for Efficient Meningitis Identification using CSF Dataset

This study introduces a novel CNN-SVM hybrid model for accurate meningitis subtype identification from Cerebrospinal Fluid (CSF) data, crucial for timely clinical decisions and improved patient outcomes. The model uniquely fuses CNNs' robust, automated feature extraction with SVMs' powerful classification, significantly enhancing diagnostic precision by reducing reliance on handcrafted features and improving generalization in complex cases. Experimental results demonstrate its superior performance, achieving 91.2% accuracy, outperforming CNN-only (85.3%) and SVM-only (81.7%) models. Furthermore, the hybrid model exhibited strong individual metrics: 90.5% Precision, 91.8% Recall, and 91.1% F1-Score. Its effectiveness and efficiency are evident, even when handling common challenges like class imbalances and biomarker variations in real-world clinical datasets. This scalable and adaptable hybrid architecture offers a promising diagnostic tool for integration into computer-aided systems, supporting clinicians in emergency settings where rapid and accurate decision-making is vital. Future work includes extending this framework to incorporate additional diagnostic modalities, such as imaging and genomic data, to foster a more comprehensive understanding of meningitis pathology.