Abstract Traditional Chinese medicine (TCM) auscultation has a long history, and with advancements in equipment and analytical methods, the quantitative analysis of auscultation parameters has determined. However, the complexity and diversity of auscultation, along with variations in devices, analytical methods, and applications, bring challenges to its standardization and deeper application. This review presents the advancements in auscultation equipment and systems, auscultation characteristic parameters, and their application in the diagnosis of pulmonary diseases and syndromes over the past 10 years, while also exploring the progress and challenges of current digital research of auscultation. This review also proposes the establishment of standardized protocols for the collection and analysis of auscultation data, the incorporation of advanced artificial intelligence (AI) auscultation analysis methods, and an exploration of the diagnostic utility of auscultatory features in pulmonary diseases and syndromes, so as to provide more precise decision support for intelligent diagnosis of pulmonary diseases and syndromes

ObjectiveTo explore the distribution pattern of tongue image characteristics in patients with glucolipid metabolic disorders and its main syndromes. MethodsA total of 841 patients with glucolipid metabolic disorders （disease group）， and 380 healthy subjects （control group） were included. The disease group was classified into three syndrome types： 283 cases of liver depression and spleen deficiency syndrome， 311 cases of phlegm-dampness obstruction syndrome， and 247 cases of qi stagnation and blood stasis syndrome. Tongue image data were collected using the TFDA-1 Tongue Diagnosis Instrument， and the TDAS V3.0 software was used to analyze the color， texture， and morphological features of the tongue body （TB） and tongue coating （TC） in patents with different syndromes of disease group （including lightness （L）， red-green axis （a）， yellow-blue axis （b）， luminance （Y）， difference between red signal and brightness （Cr）， difference between blue signal and brightness （Cb）， contrast （CON）， angular second moment （ASM）， entropy （ENT）， mean value （MEAN）， tongue coating area/tongue surface area （perAll）， and tongue coating area/non-coated area （perPart））. Logistic regression analysis was conducted to identify influencing factors for different syndrome types of glucolipid metabolic disorders. ResultsThe tongue body indicators TB-L， TB-Y， and TB-Cb in the disease group were significantly higher than those in the control group， while TB-a， TB-b， and TB-Cr were significantly lower. The tongue coating indicators TC-L， TC-Y， TC-Cb， perAll， and perPart in the disease group were significantly higher than those in the control group， while TC-a， TC-b， and TC-Cr were significantly lower （P＜0.05）. Comparing with the different syndromes in disease group， the TB-L and TB-Y of the liver depression and spleen deficiency syndrome， and the phlegm-damp obstruction syndrome were higher than those of the qi stagnation and blood stasis syndrome； the TB-a and TB-Cr of the phlegm-damp obstruction syndrome were lower than those of the qi stagnation and blood stasis syndrome； the perAll of the phlegm-damp obstruction syndrome was higher than that of the qi stagnation and blood stasis syndrome （P＜0.05）. In the analysis of the morphological characteristics of tongue signs， more spotted tongue in disease group compared with control group， more teeth-marked tongue in liver depression and spleen deficiency syndrome than the other two syndromes， more greasy coating in phlegm-damp obstruction syndrome， and more stasis spots of tongue in qi stagnation and blood stasis syndrome （P＜0.05）. Logistic regression analysis identified that greasy coating， spotted tongue， stasis spots of tongue， tooth-marked tongue， perAll， and TB-Cb are the influencing factors of liver depression and spleen deficiency syndrome； greasy coating， tooth-marked tongue， TC-Cb， and TC-Cr are the influencing factors of phlegm-damp obstruction syndrome； cracked tongue， stasis spots of tongue， tooth-marked tongue， and TB-Y are the influencing factors of qi stagnation and blood stasis syndrome （P＜0.05）. ConclusionCompared to healthy individuals， patients with glycolipid metabolic disorder have darker tongue color and thicker， greasy tongue coating. Glycolipid metabolic disorder patients of liver depression and spleen deficiency syndrome exhibit a reddish tongue with finer textures and more tooth marks； patients of phlegm-damp obstruction syndrome have lighter tongue coating with a coarser texture and a higher prevalence of greasy coating； patients of qi stagnation and blood stasis syndrome display lower tongue brightness with a higher prevalence of blood stasis spots.

Objective: To develop a non-invasive predictive model for coronary artery stenosis severity based on adaptive multi-modal integration of traditional Chinese and western medicine data. Methods: Clinical indicators, echocardiographic data, traditional Chinese medicine (TCM) tongue manifestations, and facial features were collected from patients who underwent coronary computed tomography angiography (CTA) in the Cardiac Care Unit (CCU) of Shanghai Tenth People's Hospital between May 1, 2023 and May 1, 2024. An adaptive weighted multi-modal data fusion (AWMDF) model based on deep learning was constructed to predict the severity of coronary artery stenosis. The model was evaluated using metrics including accuracy, precision, recall, F1 score, and the area under the receiver operating characteristic (ROC) curve (AUC). Further performance assessment was conducted through comparisons with six ensemble machine learning methods, data ablation, model component ablation, and various decision-level fusion strategies. Results: A total of 158 patients were included in the study. The AWMDF model achieved excellent predictive performance (AUC = 0.973, accuracy = 0.937, precision = 0.937, recall = 0.929, and F1 score = 0.933). Compared with model ablation, data ablation experiments, and various traditional machine learning models, the AWMDF model demonstrated superior performance. Moreover, the adaptive weighting strategy outperformed alternative approaches, including simple weighting, averaging, voting, and fixed-weight schemes. Conclusion The AWMDF model demonstrates potential clinical value in the non-invasive prediction of coronary artery disease and could serve as a tool for clinical decision support.

Objective: To explore the feasibility of constructing a lung cancer early-warning risk model based on facial image features, providing novel insights into the early screening of lung cancer. Methods: This study included patients with pulmonary nodules diagnosed at the Physical Examination Center of Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine from November 1, 2019 to December 31, 2024, as well as patients with lung cancer diagnosed in the Oncology Departments of Yueyang Hospital of Integrated Traditional Chinese and Western Medicine and Longhua Hospital during the same period. The facial image information of patients with pulmonary nodules and lung cancer was collected using the TFDA-1 tongue and facial diagnosis instrument, and the facial diagnosis features were extracted from it by deep learning technology. Statistical analysis was conducted on the objective facial diagnosis characteristics of the two groups of participants to explore the differences in their facial image characteristics, and the least absolute shrinkage and selection operator (LASSO) regression was used to screen the characteristic variables. Based on the screened feature variables, four machine learning methods: random forest, logistic regression, support vector machine (SVM), and gradient boosting decision tree (GBDT) were used to establish lung cancer classification models independently. Meanwhile, the model performance was evaluated by indicators such as sensitivity, specificity, F1 score, precision, accuracy, the area under the receiver operating characteristic (ROC) curve (AUC), and the area under the precision-recall curve (AP). Results: A total of 1 275 patients with pulmonary nodules and 1 623 patients with lung cancer were included in this study. After propensity score matching (PSM) to adjust for gender and age, 535 patients were finally included in the pulmonary nodule group and the lung cancer group, respectively. There were significant differences in multiple color space metrics (such as R, G, B, V, L, a, b, Cr, H, Y, and Cb) and texture metrics [such as gray-levcl co-occurrence matrix (GLCM)-contrast (CON) and GLCM-inverse different moment (IDM)] between the two groups of individuals with pulmonary nodules and lung cancer (P < 0.05). To construct a classification model, LASSO regression was used to select 63 key features from the initial 136 facial features. Based on this feature set, the SVM model demonstrated the best performance after 10-fold stratified cross-validation. The model achieved an average AUC of 0.8729 and average accuracy of 0.799 0 on the internal test set. Further validation on an independent test set confirmed the model’s robust performance (AUC = 0.823 3, accuracy = 0.729 0), indicating its good generalization ability. Feature importance analysis demonstrated that color space indicators and the whole/lip Cr components (including color-B-0, wholecolor-Cr, and lipcolor-Cr) were the core factors in the model’s classification decisions, while texture indicators [GLCM-angular second moment (ASM)_2, GLCM-IDM_1, GLCM-CON_1, GLCM-entropy (ENT)_2] played an important auxiliary role. Conclusion The facial image features of patients with lung cancer and pulmonary nodules show significant differences in color and texture characteristics in multiple areas. The various models constructed based on facial image features all demonstrate good performance, indicating that facial image features can serve as potential biomarkers for lung cancer risk prediction, providing a non-invasive and feasible new approach for early lung cancer screening.

Syndrome differentiation and treatment is a traditional Chinese medicine （TCM）-featured concept and method in diagnosis and treatment， which needs to be combined with the intelligent diagnosis of TCM in the future. At present， the intelligent diagnosis of TCM has gradually changed from the simple data-driven primary intelligence to the knowledge-driven advanced intelligence that integrates professional knowledge. Although syndrome element differentiation breaks down the elements of TCM diagnostic knowledge to form syndrome elements， which makes the original vague knowledge information more specific， the relationship between syndrome elements is not well classified and sorted out， resulting in the lack of hierarchical relationship and connection between syndrome elements， and thereby causing obstacles to the application of intelligent technology. Based on the understanding of etiology and pathogenesis from TCM， this paper concretized the syndrome element set of etiology， disease location and disease nature in TCM syndrome differentiation and treatment， and formed a knowledge framework system of syndrome elements with clear logical relationship， clear hierarchy and good explanation， thereby providing certain reference for realizing knowledge-driven advanced intelligence syndrome differentiation in the future.

Methods: From September 3, 2018, to March 23, 2024, participants with essential hypertension (receiving antihypertensive medication treatment, hypertension group) and normal blood pressure (control group) were recruited from the Cardiology Department of Shanghai Hospital of Traditional Chinese Medicine, the Coronary Care Unit of Shanghai Tenth People's Hospital, the Physical Examination Center of Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine, and the Gaohang Community Health Service Center. This study employed the propensity score matching (PSM) method to reduce study participants selection bias. Spectral information in the facial visible light spectrum of the subjects was collected using a flame spectrometer, and the spectral chromaticity values were calculated using the equal-interval wavelength method. The study analyzed the differences in spectral reflectance across various facial regions, including the entire face, forehead, glabella, nose, jaw, left and right zygomatic regions, left and right cheek regions as well as differences in parameters within the Lab color space between the two subject groups. Feature selection was conducted using least absolute shrinkage and selection operator (LASSO) regression, followed by the application of various machine learning algorithms, including logistic regression (LR), support vector machine (SVM), random forest (RF), Naïve Bayes (NB), and eXtreme Gradient Boosting (XGB). The reduced-dimensional dataset was split in a 7 : 3 ratio to establish a classification and assessment model for facial coloration related to primary hypertension. Additionally, model fusion techniques were applied to enhance the predictive power. The performance of the models was evaluated using metrics including the area under the curve (AUC) and accuracy. Shapley Additive exPlanations (SHAP) was used to interpret the outcomes of the models. Results: A total of 114 participants were included in both hypertension and control groups. Reflectance analysis across the entire face and eight predefined areas revealed that the hypertensive group exhibited significantly higher reflectance of corresponding color light in the blue-violet region (P < 0.05) and a lower reflectance in the red region (P < 0.05) compared with control group. Analysis of Lab color space parameters across the entire face and eight predefined areas showed that hypertensive group had significantly lower a and b values than control group (P < 0.05). LASSO regression analysis identified a total of 18 facial color features that were highly correlated with hypertension, including the a values of the chin and the right cheek, the reflectance at 380 nm and at 780 nm of the forehead. The results of the multi-model classification showed that the RF classification model was the most effective, with an AUC of 0.74 and an accuracy of 0.77. The combined model of RF + LR + SVM outperformed a single model in their classification performance, achieving an AUC of 0.80 and an accuracy of 0.76. SHAP model visualization results indicated that the top three contributors to ideal prediction results based on the characteristics from the facial spectrum were the reflectance at 380 nm across the entire face and of the nose as well as the a value of the chin. Conclusion Within the same age group, patients with essential hypertension exhibited significant and regular changes in facial color and facial spectral reflectance parameters after the administration of antihypertensive drugs. Furthermore, facial reflectance indicators, such as the overall reflectance at 380 nm and the a value of the chin, could offer valuable references for clinically assessing the drug efficacy and health status of patients with essential hypertension.

Methods: Data of TCM indicators (tongue, facial, and pulse diagnostics) and clinical parameters from patients diagnosed with CHD at the Cardiology Department of Shanghai Baoshan Hospital of Integrated Traditional Chinese and Western Medicine, from October 3, 2023 to March 15, 2024, were collected. Important variables were identified using importance screening and correlation analysis with CHD risk factors and laboratory markers. Six machine learning models including logistic regression (LR), decision tree (DT), support vector machine (SVM), k-nearest neighbors (KNN), random forest (RF), and gradient boosting (GB), were applied to evaluate the risk of coronary artery obstruction by combining clinical and TCM data of CHD. Model performance was assessed using metrics such as accuracy, precision, and recall, with reliability validated through ten-fold cross-validation. Results: A total of 288 patients were included in the study. Fifteen clinical risk factors, including body mass index (BMI), myoglobin, and alcohol consumption history, were incorporated into the diagnostic models. The KNN model showed good performance when combining clinical data with tongue and facial data. The SVM model performed well when clinical data was combined with pulse data. Among all the models, the KNN model with 10-fold cross-validation, which integrates the three types of TCM diagnostic data (tongue, face, and pulse) with clinical data, performs the best (accuracy: 0.837, precision: 0.814, and recall: 0.809). Conclusion Incorporating TCM diagnostic data can enhance the accuracy of coronary artery obstruction risk assessment. The KNN prediction model that integrate tongue, facial, and pulse data performs the best and can be recommended as a clinical decision support tool.

The modernization research of TCM tongue diagnosis involves a complex process of acquiring tongue diagnosis images of diseases and syndromes through machine vision and image processing technology,and conducting intelligent diagnosis and analysis.The development of artificial intelligence technology has brought new opportunities for the intelligent research of tongue diagnosis images in TCM.This article systematically reviewed the current status of the development and standard application of TCM tongue diagnosis instruments in the past decade,discussed the latest research progress in deep learning methods including tongue image quality assessment,tongue image segmentation,tongue color classification,tongue image pattern recognition,three-dimensional tongue image reconstruction,and clinical applications.It deeply analyzed the bottleneck problems existing in the current artificial intelligence tongue diagnosis technology,proposed to focus on clinical disease-syndrome combined applications,multi-level expansion,cross-modal fusion to enhance the depth and breadth of tongue diagnosis features,utilize general artificial intelligence methods to enhance the intelligence diagnosis technology of tongue images,innovate and construct a new model of intelligent tongue diagnosis with full-domain information perception,health assessment,and disease-syndrome diagnosis,in order to promote the leapfrog development of intelligent TCM diagnosis and treatment technology.

Objective To analyze the objective tongue diagnosis data characteristics of benign and malignant pulmonary nodules and to establish a lung cancer risk warning model based on the logistic regression method.Methods From July 2020 to March 2022,263 lung cancer patients(lung cancer group)from the Oncology Department of Longhua Hospital Affiliated to Shanghai University of Traditional Chinese Medicine,292 benign pulmonary nodules patients(benign pulmonary nodules group)from the Physical Examination Center of Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine,and 307 healthy individuals(healthy control group)were selected.TFDA-1 digital tongue diagnostic instrument was used to collect tongue images.Objective diagnostic features of the tongue were obtained through feature extraction technology.The distribution characteristics of the tongue indicators of the three groups of subjects were analyzed.A lung cancer warning model was established based on logistic regression method after feature screening,and the performance of the model was evaluated using sensitivity,specificity,accuracy,and AUC.Results The tongue features of patients in benign pulmonary nodules group were similar to those of the healthy control group,while the tongue features of the lung cancer group differed greatly from those of the healthy control group and benign pulmonary nodules group.The tongue features of lung cancer patients were dark and opaque,the tongue body was reddish,and the tongue coating is thin and yellowish with a greasy texture.The accuracy,sensitivity,specificity and AUC of the lung cancer warning model based on tongue image data were 70.09%,69.94%,70.29%and 0.769,respectively.After adding baseline information to the tongue image data set,the models'performance was improved.The accuracy,sensitivity,specificity and AUC of the new model based on tongue and baseline were 77.30%,75.94%,79.15%and 0.812,respectively.Conclusion The statistical characteristics of objective tongue image data between benign pulmonary nodules and lung cancer patients show significant differences.The lung cancer classification model based on objective tongue data performs well,and the objective tongue diagnosis data in TCM can provide reference for the differential diagnosis of benign pulmonary nodules and lung cancer.

Interpreting normal or diseased facial color accurately is the prerequisite for the study of facial color diagnosis technology.However,due to the variety of facial color in clinical practice and physicians may have different understanding as well as interpretation methods of the"five colors",the total rate of facial color interpretation consistency is lower than expected.The application of TCM diagnosis and treatment technology under the background of artificial intelligence puts forward higher requirements for the accuracy of facial color interpretation.This article elucidated the internal logical relationship of"the color of normal status","the color of life"and"the color of death"in Su Wen·Wu Zang Sheng Cheng Pian,explored the facial color interpretation standards,simulated the visual characteristics of"five colors"and designed the standard process of facial color interpretation.Then,the article discussed the limitations of the facial color theory in TCM,and put forward the point of view that the only way for clinical application of facial color diagnosis technology is to establish the expert consensus or guidelines that meet the needs of modern clinical diagnosis and treatment.