Chest X-ray (CXR) imaging is acknowledged as a widely used and cost-effective diagnostic test capable of identifying a broad spectrum of diseases. Despite its prevalence, diagnosing diseases accurately from CXR samples remains a challenge even for experienced radiologists. The study specifically targets pulmonary disease prediction, indicating a specialization within the broader medical context. Wiener filtering is introduced as a technique applied to the input image to eliminate noise, enhancing the quality of the image for subsequent analysis. Co-occurrence matrix features are extracted from the processed image, providing a set of statistical measures capturing spatial relationships between pixels. To increase prediction accuracy, the study employs the bio-geographical optimization algorithm for clustering. Training is conducted on the clustered images to enhance the model’s ability to discern patterns and features. The clustered image and co-occurrence matrix features extracted during the process is the novelty utilized for training a neural network. Neural networks are powerful machine learning models capable of learning complex patterns and relationships from data. Experiments are carried out on both binary and multiclass image datasets. The results indicate that the proposed model outperforms existing models, showing improvements in evaluation parameter values. Evaluation parameters could include metrics like accuracy, precision, recall, and F1 score, depending on the specifics of the study.