Search for a command to run...
In Traditional Chinese Medicine (TCM), syndrome is primarily categorized into Zang and Fu organ syndromes. Zang organs include the heart, liver, spleen, lungs, and kidneys, and diagnosis is derived from patient complaints, symptoms, and examination findings. However, Zang organ syndrome determination is highly dependent on clinical experience, resulting in diagnostic subjectivity and variability among practitioners. In addition, symptom overlap across different organs further complicates the diagnostic process and challenges standardization. To address these issues, this study proposes an artificial intelligence (AI)-based intelligent system to assist Zang organ syndrome diagnosis and provide acupuncture therapy recommendations. The system applies machine learning to improve diagnostic accuracy and consistency, utilizing the Extreme Gradient Boosting (XGBoost) algorithm due to its effectiveness in classifying complex and nonlinear patterns commonly found in TCM diagnostics. The dataset was constructed based on standardized TCM references and expert-defined symptom–syndrome relationships, consisting of 39 symptom indicators, 72 Zang organ syndrome classes, and 15,532 data instances. To address class imbalance, the Synthetic Minority Oversampling Technique (SMOTE) was applied during training. Experimental results show that the proposed model achieved an accuracy of 98.6% and a macro-average F1-score of 83% for both syndrome diagnosis and therapy recommendation tasks. The results demonstrate that the proposed AI approach effectively supports Zang organ syndrome identification and acupuncture point recommendation, contributing to improved diagnostic efficiency and decision support in TCM practice.