Similarity-based searching in multi-parameter time series databases

Abstract

We present a similarity-based searching and pattern matching algorithm that identifies time series data with similar temporal dynamics in large-scale, multi-parameter databases. We represent time series segments by feature vectors that reflect the dynamical patterns of single and multi-dimensional physiological time series. Features include regression slopes at varying time scales, maximum transient changes, auto-correlation coefficients of individual signals, and cross correlations among multiple signals. We model the dynamical patterns with a Gaussian mixture model (GMM) learned with the Expectation Maximization algorithm, and compute similarity between segments as Mahalanobis distances. We evaluate the use of our algorithm in three applications: search-by-example based data retrieval, event classification, and forecasting, using synthetic and real physiologic time series from a variety of sources.