Tian Pengfei, Chen Zonglin, Fang Biaojie, Wang Xintian, Yu Xin, Lu Mingdian
Department of General Surgery, The First Affiliated Hospital of Anhui Medical University, Hefei, 230022, China.
Department of General Surgery, The First Affiliated Hospital of Anhui Medical University, Hefei, 230022, China; Anhui University of Chinese Medicine, Hefei, 230022, China.
Eur J Surg Oncol. 2025 Jul 26;51(10):110352. doi: 10.1016/j.ejso.2025.110352.
Gastric cancer (GC) remains a major cause of cancer-related mortality, particularly in China, where early detection is hindered by reliance on invasive, resource-intensive methods like gastroscopy. This study aimed to develop a non-invasive diagnostic tool integrating tongue features derived from traditional Chinese medicine (TCM) with machine learning (ML) to enhance early GC detection.
A prospective, propensity score-matched cohort of 292 participants (146 GC, 146 non-GC) was analyzed. Standardized protocols captured tongue features (color, morphology, coating) alongside gastroscopic findings. Seven ML algorithms, including GBDT, LightGBM, and XGBoost, were trained on multimodal clinical and imaging data. Feature selection was performed using LASSO regression, and model performance was evaluated through stratified 5-fold cross-validation and a 30 % independent test set. Association rule mining (FP-Growth) was employed to explore predictive tongue-gastroscopy patterns.
GC patients exhibited more frequent bluish-purple (42 %), cracked (87 %), swollen (86 %), and prickly tongues (67 %), as well as grayish-black coatings (29 %). Non-GC individuals more often showed pale white tongues (40 %) and peeled coatings (71 %). FP-Growth identified combinations such as cracked tongue, grayish-black, and thin coatings as predictors of hemorrhagic GC (confidence = 88.89 %). LASSO highlighted key predictors including prickly tongue and CA19-9. Among the models, GBDT achieved the best performance (test AUC = 0.980, F1 = 0.932). SHAP analysis confirmed the predictive value of both tongue features and tumor markers.
TCM-based tongue diagnostics combined with ML provides a promising, non-invasive tool for early GC detection.
胃癌（GC）仍然是癌症相关死亡的主要原因，在中国尤其如此，在该国，依赖诸如胃镜检查等侵入性、资源密集型方法阻碍了早期检测。本研究旨在开发一种将源自中医（TCM）的舌象特征与机器学习（ML）相结合的非侵入性诊断工具，以加强胃癌的早期检测。
对292名参与者（146例胃癌患者、146例非胃癌患者）的前瞻性、倾向评分匹配队列进行了分析。标准化方案采集了舌象特征（颜色、形态、舌苔）以及胃镜检查结果。包括梯度提升决策树（GBDT）、轻量级梯度提升机（LightGBM）和极端梯度提升（XGBoost）在内的七种机器学习算法在多模态临床和影像数据上进行了训练。使用套索回归进行特征选择，并通过分层五折交叉验证和30%的独立测试集评估模型性能。采用关联规则挖掘（FP-增长算法）来探索预测性舌象-胃镜检查模式。
胃癌患者出现蓝紫色（42%）、裂纹舌（87%）、肿胀舌（86%）、芒刺舌（67%）以及灰黑苔（29%）的频率更高。非胃癌个体更常出现淡白舌（40%）和剥苔（71%）。FP-增长算法确定裂纹舌、灰黑苔和薄苔等组合为出血性胃癌的预测指标（置信度=88.89%）。套索回归突出了关键预测指标，包括芒刺舌和糖类抗原19-9（CA19-9）。在这些模型中，梯度提升决策树表现最佳（测试集曲线下面积[AUC]=0.980，F1值=0.932）。SHAP分析证实了舌象特征和肿瘤标志物的预测价值。
基于中医的舌诊结合机器学习为胃癌早期检测提供了一种有前景的非侵入性工具。
