ECG Prediction based on Bidirectional Time Series Chain Discovery Algorithm
Xiangwei Zheng, Xiunan Zou, Xiuxiu Ren, Cun Ji, Mingzhe Zhang · 2021
Time series chains are a set of subsequence patterns arranged in chronological order, such that each pattern is similar to the one before it, but the first and last patterns can be arbitrarily different. At present, time series chain discovery algorithms have been widely applied in various fields. However, the existing algorithms are difficult to define initial features and have high time complexity. To overcome these limitations, we propose a shapelet-based bidirectionally time series chain discovery algorithm, which adopts shapelet as initial pattern to bidirectional discover time series chain. Case study on ECG dataset shows that the proposed algorithm can reveal the dynamic evolution of time series. At the same time, we also present the case study on time series prediction. Experimental results verify the effectiveness of the proposed algorithm, meanwhile, the algorithm can provide useful insights for heart disease patients.