EventDTW: An Improved Dynamic Time Warping Algorithm for Aligning Biomedical Signals of Nonuniform Sampling Frequencies

Yuankai Qi,Jeffrey Olgin,Robert Avram,Will Ke Wang,Brinnae Bent,Yihang Jiang,Jessilyn Dunn

doi:10.3390/s20092700

Yuankai Qi, Jeffrey Olgin + Show 5 more

Open Access

https://doi.org/10.3390/s20092700

Copy DOI

Journal: Sensors	Publication Date: May 9, 2020
Citations: 32	License type: CC BY 4.0

Affiliation: Duke University, University of California, San Francisco

Abstract

The dynamic time warping (DTW) algorithm is widely used in pattern matching and sequence alignment tasks, including speech recognition and time series clustering. However, DTW algorithms perform poorly when aligning sequences of uneven sampling frequencies. This makes it difficult to apply DTW to practical problems, such as aligning signals that are recorded simultaneously by sensors with different, uneven, and dynamic sampling frequencies. As multi-modal sensing technologies become increasingly popular, it is necessary to develop methods for high quality alignment of such signals. Here we propose a DTW algorithm called EventDTW which uses information propagated from defined events as basis for path matching and hence sequence alignment. We have developed two metrics, the error rate (ER) and the singularity score (SS), to define and evaluate alignment quality and to enable comparison of performance across DTW algorithms. We demonstrate the utility of these metrics on 84 publicly-available signals in addition to our own multi-modal biomedical signals. EventDTW outperformed existing DTW algorithms for optimal alignment of signals with different sampling frequencies in 37% of artificial signal alignment tasks and 76% of real-world signal alignment tasks.

Highlights

Dynamic time warping (DTW) is a class of pattern matching algorithms for aligning temporal sequences
The error rate compares the alignment resulting from using a dynamic time warping (DTW) algorithm against a perfect alignment that we calculated in II.B, and outputs a proportion score of error
We tested the performance of DTW, dDTW, shape DTW (sDTW), and eDTW using singularity score (SS) and error rate (ER) on a goldstandard open-access public dataset, including 84 different signals collected from various scenarios

Summary

Introduction

Dynamic time warping (DTW) is a class of pattern matching algorithms for aligning temporal sequences. DTW aims to provide an optimal similarity-based match between two sequences regardless of dynamic spatiotemporal differences, searching for the optimal path by minimizing cumulative point-to-point distances. DTW allows for a flexible definition of cumulative point-to-point distances to optimize alignment performance for different datasets. Because of this flexibility, DTW is widely used in speech recognition under varying speaking speeds [1], gesture recognition [2,3], and time series clustering [4]. Measurement modalities utilizing multiple simultaneous sensors with different sampling frequencies and/or uneven sampling rates are common in many applications, in biomedicine [5,6]. We have observed that DTW performs poorly when applied to signals

Methods

Results

Discussion

Conclusion