Feature Selection and Feature Extraction-Aided Classification Approaches for Disease Diagnosis

In this paper, the application of machine learning approach in the construction of disease diagnosis system is introduced. So as to introduce machine learning approach into clinical medicine, and the performance of machine learning medical diagnosis model with different feature extraction or selection approaches are studied. The respective advantages of feature selection and feature extraction are employed to eliminate redundant information. Firstly, a feature extraction algorithm based on partial least squares is designed and proposed, and compared with the typical feature extraction approach principal component analysis. Then partial least square approach and recursive feature elimination are used to analyze the correlation of the original variable set, so as to analyze the possibility of the disease caused by the original disease variables collected. Finally, the proposed approach is verified on a cervical cancer data. Experimental results indicate that the proposed approach can achieve better feature extraction effect for cervical cancer diagnosis, which can improve the diagnosis accuracy on the premise of eliminating redundant information. In the aspect of correlation analysis of pathogenic factors, experimental results show that recursive feature elimination has better effect.