Correlation Analysis between Syndrome of TCM and Clinical Features and AI Analysis of Local Features of"Laryngeal Cough"
Objective:To analyze the correlation between clinical symptoms and traditional Chinese medicine(TCM)syndromes of"laryngeal cough"and explore the application of artificial intelligence(Al)in the classification of pharyngeal image features,so as to improve the objective information collection of clinical symptoms,expand the scope of TCM inspection,and service for improving the systemic syndrome differentiation by the refined local syndrome differentiation.Methods:142 patients with laryngeal cough who were admitted to the outpatient department of Yueyang Hospital of Integrated Traditional Chinese and Western Medicine from August 2022 to February 2023 were collected,and their related clinical data and endoscopic image information of the posterior pharyngeal wall were recorded,then the correlation between clinical symptoms and TCM syndromes of"laryngeal cough"were analyzed.Furthermore,classify and label all pharyngeal images and hand over them to AI for local features learning to explore the accuracy of manual classification of the local features.Results:The results of the logistic model analysis showed that correlation exists between different clinical symptoms and different TCM syndromes of"laryngeal cough".In the univariate regression model,moderate pharyngeal dryness(OR=2.630)and moderate itchy throat(OR=3.290)are positively correlated with phlegm deficiency syndrome;Severe pharyngeal dryness(OR=5.786),moderate pharyngeal mucosal congestion(OR=7.360)and moderate lymphoid follicular hyperplasia of posterior pharyngeal wall(OR=4.200)are positively correlated with fire-heat syndrome;severe abnormal sensation of throat(OR=5.353)and severe pharyngeal mucosal congestion(OR=5.194)are positively correlated with blood stasis syndrome;the remaining symptoms did not correlate significantly with any of the three syndromes.Multi-factor logistic regression analysis of the meaningful results yielded:relative to no itchy throat,moderate itchy throat increased the risk of phlegm evidence,and the difference was statistically significant[OR=3.269,95％CI(1.060,10.079),P=0.039];relative to no dry throat,moderate dry throat increased the risk of phlegm evidence,and the difference was statistically significant[OR=2.649,95％CI(1.041,6.745),P=0.041];relative to no dry throat,severe dry throat increased the risk of fire-heat evidence,and the difference was statistically significant[OR=24.615,95％CI(2.774,218.403),P=0.004];relative to the absence of pharyngeal mucosal congestion,moderate pharyngeal mucosal congestion increased the risk of fiery evidence,and the difference was statistically significant[OR=21.819,95％CI(3.526,135.027),P=0.001];relative to mild posterior pharyngeal wall lymphoid follicular hyperplasia,moderate posterior pharyngeal wall lymphoid follicular hyperplasia increased the risk of pyrogenic evidence,and the difference was statistically significant[OR=11.657,95％CI(1.942,69.99),P=0.007];however there was no significant significance in pharyngeal foreign body sensation,pharyngeal mucosal congestion and blood stasis.AI analysis showed that the overall classification accuracy was 74.8％.Conclusion:Some clinical features of laryngeal cough are correlated with different TCM syndromes which could be used as a reference for systemic syndrome differentiation;the others are not necessarily correspond to the systemic condition,and we could treate the laryngeal cough from a combination of systemic syndrome differentiation and local syndrome differentiation;the combination of AI image analysis technology and extended inspection could broaden the thinking of syndrome differentiation and provide reference for fine syndrome differentiation.
laryngeal coughsyndrome of TCMlocal featurecorrelation analysisartificial intelligence
万方数据
维普
