The athletes' oral odors are directly related with physiological factors such as large amount of exercise and mental factors such as mental stress. As a kind of human body odor, oral odors have a series of characteristics such as stability, uniqueness, simplicity of feature extraction and convenience, which can be used as the basis of disease diagnosis. Based on the biological characteristic analysis technology, this study analyzes the extraction and detection techniques of athletes' oral odors. The stability of exhaled gas of individuals is verified by odor chromatography. The results show that the similarity of exhaled gas chromatograms of the three groups in 30 days keeps about 88% on average, which indicates that the oral odors of the athletes remain stable. In the electronic nose system, the Linear Discriminant Analysis and Principal Component Analysis are used to classify diseases associated with oral odors. The results show that Linear Discriminant Analysis is superior to Principal Component Analysis in classifying patients with diseases and healthy people, which provides technical support and theoretical basis for exploring athletes' oral odors and their physiology, as well as its relationship with mental problems.