Enhanced Fruit Fly Optimization Algorithm for Wrapper Based Feature Selection in Cancer Prediction

Nowadays, smart healthcare appliances generate vast amount of medical data. Such huge amount of data requires classification methodologies, through which disease diagnostics can be carried out. Chronic diseases like cancer can be predicted in advance to help the patients to take treatment in initial stages of the disease. This article proposes an improvised feature selection and classification -method for cancer prediction. Here we propose an enhanced fruit fly optimization algorithm in KNN classifier for searching strategy with wrapper methods for feature selection. Lung cancer data set, breast cancer data set and leukemia data set has been taken from the UCI repository for testing the performance of the proposed model. For Lung cancer data set, accuracy, specificity, sensitivity, and fscore value observed are 96.97 %, 96.28 %, 96.35 %, and 96.34% respectively. For breast cancer dataset, accuracy, specificity, sensitivity, and fscore value observed are 99.97 %, 98.2 %, 96.85 %, and 90.35% respectively. For Breast cancer dataset, accuracy, specificity, sensitivity, and fscore value observed are 96.67%, 98.28 %, 95.85 %, and 93.85% respectively. The proposed method works well with various dataset and shows prominent accuracy improvement over the existing systems.