A novel trigonometric subpopulation and sine cosine range optimization framework to classify Helicobacter pylori infection in a south Indian cross-sectional study

Abstract Helicobacter pylori (H. pylori) Infection is considered a major cause of gastrointestinal illnesses like peptic ulcer and gastric cancer. Early diagnosis of H. pylori infection will reduce the risk of disease progression. The convolutional neural network (CNN) model, when combined with the proposed TSOA, achieved 90.7% ± 0.028 accuracy (ACC). Also, 90.7% ± 0.028 sensitivity (SENS), 94.3% ± 0.052 specificity (SPEC), with a precision (PREC) of 91.0% ± 0.028 and an F1 Score (F1) of 90.6% ± 0.027 were achieved on unseen images under tenfold cross-validation, surpassing other methods. Additionally, the results were compared with standard benchmark functions and three established optimization algorithms: the Grey Wolf Optimizer (GWO), Whale Optimization Algorithm (WOA), and Arithmetic Optimization Algorithm (AOA), to evaluate the effectiveness of our algorithm. Integrating in a real-time setting will help gastroenterologists identify H. pylori early. In the future, we plan to expand our dataset to a more generalized population so that our results can be validated using datasets from different demographic locations. We would implement explainable AI techniques, privacy-preserving techniques, and federated learning that would help in collaborations among the hospitals.