Abstract: Non-invasive detection of arteriosclerosis for the prevention of cardiovascular events is important. In order to improve the accuracy of arteriosclerosis recognition, this paper puts forward multi-source signal（ECG and pulse wave） recognition model to detect arteriosclerosis based on ant colony clustering algorithm with mutation features. The performance of method is tested by 40 sets actual samples data. The compared results between simulation and professional clustering show that the proposed method can improve the atherosclerosis recognition rate; therefore the proposed model is an effective atherosclerosis non-invasive detection model.

Key words: ECG signal, pulse wave signal, arteriosclerosis, ant colony algorithm, non-invasive detection

摘要： 动脉硬化无创检测对于预防心血管事件具有重要意义。然而，基于心电信号或脉搏波信号的单一特征源的无创动脉硬化检测无法全面反映心血管动脉硬化事件。为了提高动脉硬化无创检测识别精度，提出了基于心电信号、脉搏波信号的多源数据无创动脉硬化识别方法，构建了具有变异特性的蚁群聚类算法，对提取的40组临床心电、脉搏波信号的特征值向量进行监督分类。通过对系统测试结果与专家分类结果对比分析，表明该方法提高了单一特征源的动脉硬化识别率，是一种有效的动脉硬化无创识别方法。

关键词: 心电信号, 脉搏波信号, 动脉硬化, 蚁群算法, 无创检测