Robust filtering of time series with trends

We develop and test a robust procedure for extracting an underlying signal in form of a time-varying trend from very noisy time series. The application we have in mind is online monitoring data measured in intensive care, where we find periods of relative constancy, slow monotonic trends, level shifts and many measurement artifacts. A method is needed which allows a fast and reliable denoising of the data and which distinguishes artifacts from clinically relevant changes in the patient's condition. We use robust regression functionals for local approximation of the trend in a moving time window. For further improving the robustness of the procedure, we investigate online outlier replacement by e.g. trimming or winsorization based on robust scale estimators. The performance of several versions of the procedure is compared in important data situations and applications to real and simulated data are given.

[1]  Ursula Gather,et al.  Robust Estimation of Scale for Local Linear Temporal Trends , 2002 .

[2]  Roland Fried,et al.  Computing the update of the repeated median regression line in linear time , 2003, Inf. Process. Lett..

[3]  A F Smith,et al.  Monitoring renal transplants: an application of the multiprocess Kalman filter. , 1983, Biometrics.

[4]  R. Martin,et al.  Bias Robust Estimation of Scale , 1993 .

[5]  R. Zamar,et al.  Global robustness of location and dispersion estimates , 1999 .

[6]  Michèle Basseville,et al.  Detection of abrupt changes: theory and application , 1993 .

[7]  M. Imhoff,et al.  Statistical pattern detection in univariate time series of intensive care on-line monitoring data , 1998, Intensive Care Medicine.

[8]  A. Siegel.  Robust regression using repeated medians , 1982 .

[9]  John W. Tukey,et al.  Exploratory Data Analysis. , 1979 .

[10]  Jianqing Fan,et al.  Robust Non-parametric Function Estimation , 1994 .

[11]  R. Grübel.  The Length of the Shorth , 1988 .

[12]  Ursula Gather,et al.  Robust signal extraction for on-line monitoring data , 2004 .

[13]  Fred Godtliebsen,et al.  M-Smoother with local Linear Fit , 2002 .

[14]  S. Sheather,et al.  A Cautionary Note on the Method of Least Median Squares , 1992 .

[15]  P. Rousseeuw.  Least Median of Squares Regression , 1984 .

[16]  Ursula Gather,et al.  Online monitoring of high dimensional physiological time series: A case study , 2001 .

[17]  P. Rousseeuw,et al.  A robust scale estimator based on the shortest half , 1988 .

[18]  P. Rousseeuw,et al.  Alternatives to the Median Absolute Deviation , 1993 .

[19]  Ursula Gather,et al.  Fast and Robust Filtering of Time Series with Trends , 2002, COMPSTAT.

[20]  Peter J. Rousseeuw,et al.  Time-Efficient Algorithms for Two Highly Robust Estimators of Scale , 1992 .

[21]  P. L. Davies.  Aspects of Robust Linear Regression , 1993 .