人工智能在中医诊断中的应用进展

本研究立足于各类人工智能算法的数学原理，阐述了人工智能在中医诊断中的应用现状及问题。其中传统机器学习算法，如支持向量机、贝叶斯算法等因其小样本学习的特性，在闻诊、问诊等场景具备较高的精度与稳健性；而近年来新兴的深度学习算法则可以处理如图像、音频信号、文本等非结构化数据，与望诊、切诊等场景相契合；多模态深度学习则可以充分挖掘望闻问切数据中的信息，并在特征空间中进行隐式的四诊合参。人工智能的引入可以进一步推动中医的客观化、定量化发展，但其数据驱动的特性要求进一步规范现行的中医数据库建立流程。

Advances in the application of artificial intelligence in traditional Chinese medicine diagnosis

Abstract: Based on the mathematical principles of various artificial intelligence algorithms, the current situation and problems of artificial intelligence application in traditional Chinese medicine (TCM) diagnosis is expounded. The traditional machine learning algorithms, such as support vector machines and Bayesian algorithms, have high accuracy and robustness in auscultation, inquiry and other scenarios because of their characteristic of small sample learning. Some deep learning algorithms emerging in recent years which can process unstructured data such as images, audio signals, texts, etc. are suitable for scenarios such as inspection and palpation. Multi-modal deep learning can fully mine the information in the data of inspection, auscultation, inquiry and palpation, and perform implicit analysis of 4 TCM diagnostic methods in the feature space. The introduction of artificial intelligence can further promote the objective and quantitative development of TCM, but its data-driven nature requires further standardization of the current TCM database establishment.