In this paper, a strategy for data-driven interpretation of time series based on a formalization of empirical strategies is proposed. It includes the systematic extraction of interpretable features, supervised and unsupervised learning techniques. It is applied to the interpretation of time series from clinical motion analysis for patients with an incomplete spinal cord injury.
Print ISSN: 0178-2312
Volume: 52, 08/2004
Pages: 359 - 369