Alcoholism Detection via Wavelet Energy and Logistic Regression

摘要

In this study, we proposed an application of alcoholism detection via wavelet energy and logistic regression. We collected data sets of 70 volunteers who signed up through advertising, among which 35 were with alcoholism and the rest were healthy. We first used wavelet energy (WN) to extract brain images features. Then, we employed logistic regression (LR) as the classification tool. Finally, we used 5-fold stratified cross validation to verify classifier performance. Our method achieves a sensitivity of 84.00±3.86%, a specificity of 84.86±3.03%, and an accuracy of 84.43±1.42%. Our method gives better performance than HWT and ANN-GA.