Cancer prediction in the early stage is a topic of major interest in medicine since it allows accurate and efficient actions for successful medical treatments of cancer. Mostly cancer datasets contain various gene expression levels as features with less samples, so firstly there is a need to eliminate similar features to permit faster convergence rate of classification algorithms. These features (genes) enable us to identify cancer disease, choose the best prescription to prevent cancer and discover deviations amid different techniques. To resolve this problem, we proposed a hybrid novel technique CSSMO-based gene selection for cancer classification. First, we made alteration of the fitness of spider monkey optimization (SMO) with cuckoo search algorithm (CSA) algorithm viz., CSSMO for feature selection, which helps to combine the benefit of both metaheuristic algorithms to discover a subset of genes which helps to predict a cancer disease in early stage. Further, to enhance the accuracy of the CSSMO algorithm, we choose a cleaning process, minimum redundancy maximum relevance (mRMR) to lessen the gene expression of cancer datasets. Next, these subsets of genes are classified using deep learning (DL) to identify different groups or classes related to a particular cancer disease. Eight different benchmark microarray gene expression datasets of cancer have been utilized to analyze the performance of the proposed approach with different evaluation matrix such as recall, precision, F1-score, and confusion matrix. The proposed gene selection method with DL achieves much better classification accuracy than other existing DL and machine learning classification models with all large gene expression dataset of cancer.

Neurotransmitter molecules play fundamental roles in neuronal communication and are closely associated with the regulation of cognitive, emotional, and motor functions. Alterations in neurotransmitter-related molecular systems are implicated in various neuropsychiatric and neurodegenerative disorders, underscoring the importance of understanding their structural and physicochemical characteristics. In this study, a graph-theoretic quantitative structure–property relationship (QSPR) framework is developed to investigate the relationships between neurotransmitter molecular structures and selected physicochemical properties using four distance-based topological descriptors: Mostar (Mo), Padmakar–Ivan (PI), Szeged (Sz), and geometric–arithmetic (GA) indices. These descriptors provide mathematically interpretable representations of molecular connectivity, distance relationships, and structural organization. Linear regression models were constructed to examine the relationships between the descriptors and five physicochemical properties, namely molecular weight (MW), polar surface area (PSA), structural complexity (C), solubility (S), and melting point (MP). Model performance was assessed using the coefficient of determination (\(R^2\)), adjusted \(R^2\), root mean square error (RMSE), and actual-versus-fitted value comparisons. The results demonstrate that model fit is strongly property– and descriptor-dependent. The PI descriptor exhibits the strongest regression relationships for MW and C, while Mo provides the strongest model fit for PSA. GA shows the highest relative model fit for solubility among the investigated descriptors, whereas Sz provides the highest relative fit for melting point, although the corresponding linear association remains weak. Overall, the distance-based descriptors provide substantially stronger linear relationships with MW, PSA, and C than with S and MP, indicating that these latter properties are governed by additional structural, conformational, and physicochemical factors that cannot be adequately captured by a single two-dimensional topological descriptor. The findings demonstrate the utility of graph-theoretic descriptors as interpretable tools for preliminary QSPR relationship analysis and provide a basis for future studies incorporating larger and chemically diverse datasets, complementary molecular descriptors, and independent external validation.

Cancer prediction in the early stage is a topic of major interest in medicine since it allows accurate and efficient actions for successful medical treatments of cancer. Mostly cancer datasets contain various gene expression levels as features with less samples, so firstly there is a need to eliminate similar features to permit faster convergence rate of prediction algorithms. These features (genes) enable us to identify cancer disease, choose the best prescription to prevent cancer and discover deviations amid different techniques. To resolve this problem, we proposed a hybrid novel technique CSSMO-based gene selection for cancer prediction. First, we made alteration of the fitness of Spider Monkey Optimization (SMO) with Cuckoo search (CS) algorithm viz., CSSMO for feature selection, which helps to combine the benefit of both metaheuristic algorithms to discover a subset of genes which helps to predict a cancer disease in early stage. Further, to enhance the accuracy of the CSSMO algorithm, we choose a cleaning process, minimum redundancy maximum relevance (mRMR) to lessen the gene expression of cancer datasets. Next, these subsets of genes are classified using Deep Learning (DL) to identify different groups or classes related to a particular cancer disease. Six different datasets have been utilized to analyze the performance of the proposed approach in terms of cancer sample classification and prediction with Recall, Precision, F1-Score, and confusion matrix. The proposed gene selection method with DL achieves much better prediction accuracy than other existing Deep Learning (DL) and Machine learning models with large gene expression dataset.