Integrating HRPGW optimization with ALSTM for enhanced cancer diagnosis in gene expression microarray data analysis

Abstract

Cancer diagnosis using gene expression microarray data involves analysing gene expression patterns to classify samples as cancerous or non-cancerous, aiding in early detection and treatment planning for various types of cancer. Challenges in cancer diagnosis from gene expression microarray data include noise and variability in data, feature selection from high-dimensional datasets, overfitting, class imbalance, and the need for robust algorithms to effectively distinguish between cancerous and non-cancerous samples. This work involves a comprehensive approach to analysing microarray data for cancer diagnosis. It begins with the selection of relevant microarray data, followed by essential data pre-processing steps such as normalization and handling missing values to ensure data quality. Dimensionality reduction techniques, particularly Principal Component Analysis (PCA), are employed to reduce the complexity of the dataset. Feature selection is then performed using the Hybrid Red Piranha Grey Wolf (HRPGW) algorithm, which combines the optimization power of Grey Wolf Optimization with the exploration capability of Red Piranha. Subsequently, classification models Artificial Long Short-Term Memory (ALSTM), including Artificial Neural Networks (ANN) and Long Short-Term Memory Networks (LSTM), are utilized for accurate cancer diagnosis. This integrated approach ensures that the most relevant features are extracted from the data, optimizing classification performance while mitigating the effects of noise and high dimensionality inherent in microarray datasets, ultimately enhancing the accuracy and reliability of cancer diagnosis.