动态心电波形数据的聚类有效性评价方法研究

Study on clustering validity evaluation method on AECG waveform data

研究了动态心电信号的波形形态特点,提出了改进的基于品质因子(Figure Of Merit,FOM)的聚类有效性评价策略——HW-FOM(Hausdorff Weighted-Figure Of Merit)。该策略将完整的心电波形分段,并赋予不同的权重。分别计算任意两个心电波形间的对应段的Hausdorff距离,用以拟合心电波形的形态。通过汇总计算结果描述心电波形间的差异,评价动态心电波形聚类结果的有效性。通过利用MIT-BIH心律失常数据进行实验,结果表明HW-FOM方法的评价结果与实际的数据分类状况呈线性相关,适于动态心电波形的聚类有效性结果的评价。

According to the study on the feature of Ambulatory Electrocardiogram waveform（AECG）, a cluster assessment strategy HW-FOM（Hausdorff Weighted-Figure Of Merit） is presented on the base of Figure Of Merit（FOM）.The strategy divides whole ECG waveform into several segments, and assigns weights to each segment.It computes corresponding segment distance of any two ECG waveforms by Hansdorff method, which can fit data to ECG shape.It describes the difference among ECG waveform by summed result,which is used to evaluate the effective of cluster result.MIT/BIH arrhythmia data is used in the experiment,the result shows that the result of HW-FOM is linear correlation with the real classification result, and HW-FOM is suitable for cluster validity evaluation on AECG data.