"Hybrid Model for Cirrhosis Detection Using Volatile Organic Compounds (VOCs) and Kernel-Based Support Vector Machines"

"Early detection of liver cirrhosis remains a significant challenge in modern medicine. Recent studies suggest that volatile organic compounds (VOCs) present in human breath can serve as potential biomarkers for non-invasive diagnosis. This work proposes a hybrid artificial intelligence model for the detection of cirrhosis based on VOC data analysis. The proposed approach integrates a neural network trained with backpropagation for feature extraction and a kernel-based Support Vector Machine (SVM) classifier to improve class separation.A dataset of 1000 simulated VOC samples was generated, including gases such as acetone, ethanol, ammonia, methane, isoprene, and hydrogen sulfide, along with environmental parameters such as temperature and humidity. The neural network learns nonlinear relationships among the features and produces a reduced latent representation of the data. These extracted features are then used to train an SVM classifier with a radial basis function (RBF) kernel to enhance decision boundaries in a higher-dimensional feature space.To visualize the separability of the classes, Principal Component Analysis (PCA) was applied to reduce the feature space to two dimensions. The model performance was evaluated using accuracy, confusion matrix analysis, and receiver operating characteristic (ROC) curves. Experimental results show that the hybrid model achieves high classification accuracy and demonstrates the feasibility of combining deep learning feature extraction with kernel-based machine learning methods for medical breath analysis.The proposed methodology provides a promising framework for non-invasive liver disease screening systems based on breath analysis and artificial intelligence techniques."