Non-small cell lung cancer (NSCLC), primarily consisting of lung squamous cell carcinoma (LUSC) and lung adenocarcinoma (LUAD), is a significant cause of cancer-related death globally. Accurate subtyping and identifying reliable biomarkers for NSCLC are needed to design personalized therapy; however, the molecular heterogeneity and high dimensionality of gene expression data pose significant challenges. This work proposes iPSOgs, an enhanced particle swarm optimization algorithm with a golden section search refinement strategy and an adaptive crossover-based solution generation method to improve search efficiency and convergence stability. The XGBoost classifier’s hyperparameter optimization and gene selection are done simultaneously using iPSOgs to predict NSCLC subtypes. On transcriptomic datasets (TCGA-LUAD, TCGA-LUSC, and GSE81089), the developed framework performs exceptionally well, achieving an accuracy of 0.9580, a receiver operating characteristic area under the curve of 0.9879, an F1 score of 0.9560, a recall of 0.9456, and a precision of 0.9668. The robustness of iPSOgs in managing complex optimization landscapes is further confirmed by comparison with cutting-edge metaheuristics on the CEC2017 test suite. The role of the biologically significant genes DSG3, KRT5, and SPRR2E, which were confirmed as involved in NSCLC pathogenesis by enrichment and protein-protein analysis, is further highlighted by the SHAP-based explanation. This work demonstrates the efficacy of iPSOgs as a diagnostic and biomarker discovery tool, offering a scalable and explainable solution for precision oncology in lung cancer.

Background: Breast cancers are heterogeneous in nature, including many molecular subtypes, each displaying varying characteristics in clinical outcomes as well as in responses to treatments. Subtyping requires absolute precision for the application of precision medicine; however, this is not an easy task, given the dimensionality as well as noise in miRNA expression profiles. Even though miRNAs display potential as a biological marker for subtyping breast cancers, feature selection and optimizing learning algorithms would help harness their potential as a diagnostic tool. Methods: We propose the Adaptive Hill Climbing Artificial Lemming Algorithm (AHALA), a hybrid optimization framework that integrates the global search capability of the Artificial Lemming Algorithm with an adaptive hill-climbing local search strategy. Low-variance filtering and differential gene expression analysis were first applied to reduce dimensionality and enhance biological relevance. AHALA was then used to optimize deep neural network hyperparameters for miRNA-based multi-class breast cancer subtype classification. The method was validated using TCGA breast cancer miRNA expression data and benchmarked against state-of-the-art optimization algorithms using the CEC2021 test suite. Results: AHALA had a high classification performance measure for each type of breast cancer with a mean accuracy of 95.74%, precision of 95.98%, recall of 95.74%, F1 measure of 95.74%, and AUC value of 0.9682. The new algorithm had superior convergence and significance compared with other optimization algorithms. Feature selection revealed miRNAs that belong to each subtype, such as hsa-miR-190b, hsa-miR-429, hsa-miR-505-3p, hsa-miR-3614-5p, and hsa-miR-935. Conclusions: The AHALA framework offers a potent and efficient method of performing miRNA-based subtyping of breast cancer that integrates global exploration and local search to its advantage. Its high level of classification, stability, and ability to identify biologically important biomarkers mark this method as promising.