<p>Chronic obstructive pulmonary disease (COPD) is one of the major global health challenges for which appropriate diagnostic approaches need to be made accurately and effectively to enhance patient outcomes and clinical decision-making. The current work is focused on some machine learning (ML)-based methods that predict COPD, specifically the Histogram Gradient Boosting Classifier (HGBC), enhanced with optimization algorithms such as Atom Search Optimization (ASO), Wild Geese Algorithm (WGA), Giant Trevally Optimizer (GTO), and Genetic Algorithm (GA). The novelty lies in the use of such optimization algorithms for improved performance of models overcoming limitations in traditional diagnostic methods. The aggregation of patient demographics, past medical history, symptoms, physiological variables, and environmental data has been used for thorough model training and evaluation. Among them, the HGGT model was exceptionally good, with an accuracy of 0.9427, outperforming the HGWG model, which had an accuracy of 0.9140, and the HGAS model with 0.8925. These results have pointed out the superior predictive capabilities of the HGGT and HGWG models, offering a novel framework for reliable COPD diagnosis. It therefore contributes to the existing literature by illustrating how challenging tasks in the prediction of COPD are amenable to cutting-edge ML and optimization techniques that may bring forth potentially ground-breaking clinical applications supported by evidence.</p> Graphical Abstract <p></p>