GENE SELECTION AND MODIFIED LONG SHORT TERM MEMORYNETWORK-BASED LUNG CANCER CLASSIFICATION USING GENE EXPRESSION DATA

Abstract

Lung cancer is one of the fatal forms of cancer worldwide. Genetic variability has been identified as influencing a person vulnerability to lung cancer in epidemiologic research. A new study undertaken by a team of experts from the United States National Cancer Institute on 14,000 Asian women discovered that Asian women, regardless of whether they smoke or not, are more likely to acquire cancer owing to genetic abnormalities. Early detection of this lethal disease is a novel clinical application of microarray data. Recent research establishes a model for the early diagnosis of lung cancer. Additionally, multilayer perceptron, random subspace, and SMO (Sequential Minimal Optimization) approaches are used for classification. While information acquisition is typically a good indicator of an attribute significance, it is not perfect. A noticeable issue develops when knowledge gain is applied to qualities that might take on many distinct values. This paper provides an efficient gene selection model based on the Improved Whale Optimization Algorithm (IWOA) to address these concerns. It saves time and identifies relevant genes from gene expression data, increasing lung cancer categorization accuracy. Then, a Modified Long Short-Term Memory (MLSTM) Network is used to classify lung cancer. It accepts specified genes as inputs and determines which class they belong to, such as lung cancer or normal subjects. As demonstrated by empirical observations, the suggested model is effective in precision, recall, accuracy, and f –measure.

Keywords

• lung cancer
• early stage
• developing cancer
• genetic variations
• feature selection
• information gain attribute
• whale optimization