复杂网络理论在酗酒分类中的应用研究

摘要

Alcoholism has become the third largest factor affecting human health ,but pheno‐type characteristics of alcohol rolated brain damage are not obvious .At present ,the study of alco‐holics electroencephalogram(EEG) is mostly concentrated in the difference between the local fea‐tures of brain ,but rarely classified between alcoholics and control group .This study used com‐plex network theory to construct and analyze the alcoholics EEG function brain networks based on EEG data ,and dug out specific network indicators as classification features by statistical test and machine learning .The accuracy can achieve 76.8% using SVM .The results show that the properties of brain networks can be used as early objective indicators of alcoholism ,and applied to the clinical diagnosis of alcoholism disease .%在酗酒者脑电数据基础上，利用复杂网络理论构建并分析了酗酒者EEG功能脑网络，通过统计检验和机器学习算法挖掘出正常被试和酗酒被试之间的特异性网络指标作为分类特征，用SVM 分类的准确率最高可达76.8％。分类结果表明，通过复杂网络理论得到的脑网络属性可作为酗酒疾病的早期客观指标，应用到酗酒疾病的临床辅助诊断。