Speaker
Jo Røislien, Researcher, Department of Biostatistics, Institute of Basic Medical Sciences, University of Oslo and presenter of the TV-show 'Siffer'.
Abstract
Clinical studies frequently include repeated measurements on each individual, often for long periods of time. We present a methodology for extracting common temporal features across a set of individual time series observations. In particular, exploring extreme observations within the time series, i.e. spikes, as a possible common temporal phenomenon was of interest. Wavelet basis functions are attractive in this sense, as they are localized in both time and frequency domains simultaneously, allowing for localized feature extraction from a time-varying signal. We applied the following six-step procedure; missing handling by multiple imputation (MI), transformation to wavelet domain, wavelet thresholding, feature extraction in wavelet domain by principal component analysis, combination of MI results by matrix averaging, and inverse wavelet transformation to time domain for clinical interpretation. The methodology is demonstrated on a subset of a large fetal activity study aiming to identify temporal patterns in fetal movement (FM) count data in order to explore formal FM counting as a screening tool for preventing adverse birth outcomes.