Diagnosis Models Of Chronic Gastritis And Peptic Ulcer Based On Rapid Gas Chromatography Of Exhalations

Chronic gastritis and peptic ulcer are clinically common gastrointestinal diseases.Chronic gastritis increases the risk of peptic ulcer and gastric cancer,and peptic ulcer are often accompanied by hemorrhage and perforation.Chronic gastritis and peptic ulcer are insidious,long lasting,and complicated to treat.These diseases can have a significant impact on patients’ life.The current gold standard for clinical testing is endoscopy,but endoscopy examination is often accompanied by discomfort,and may cause digestive tract damage.Additionally,painless endoscopy is associated with the risk of anesthesia.In this context,the development of a painless,non-invasive test to screen for chronic gastritis and peptic ulcer may be helpful for early detection and intervention treatment of these diseases.Volatile organic compounds in human exhaled breath have been shown to be associated with gastric diseases.In this study,the exhalation data of 547 subjects are collected by a breath detection device based on rapid gas chromatography and surface acoustic wave sensor,and two diagnosis models are constructed for the assistant diagnosis of chronic gastritis and peptic ulcer based on the exhalation data.The augmented algorithm based on DTW(Dynamic Time Warping)and DBA(Dynamic Time Warping Barycenter Averaging)algorithm,and GAN(Generative Adversarial Networks)framework are used to augment the exhalation data,and convolutional neural network is used to learn the data.With these methods,two disease diagnosis models are successfully constructed.Model A can distinguish healthy subjects from patients with chronic gastritis or peptic ulcer,and model B can distinguish patients with chronic gastritis from patients with peptic ulcer.The Accuracy,Precision,Recall,F1 of model A are respectively 0.911,0.967,0.831 and 0.893.The Accuracy,Precision,Recall,F1 of model B are respectively0.670,0.649,0.793 and 0.706.The results of the study confirm the feasibility and potential clinical value of using exhaled breath to screen patients with chronic gastritis and peptic ulcer.