13th Meeting of Researchers in Biometrics/Statistics
June 19 - 22, 2006, Cavtat / Dubrovnik, Croatia

Co-organizers
HBMD Croatian Biometric Society SRCE - University Computing Centre

BIOSTAT Chair
Marija Pecina

BIOSTAT Topics
Problems in bio-information analysis

• Data Visualization
• Space Reduction Methods
• Prediction and Classification
• Time and Space Modeling
• Computationally Intensive Methods
• Other

11th SCHOOL OF BIOMETRICS

SCHOOL OF BIOMETRICS Topic

Analyzing On-line monitoring data from Intensive Care Alarm Systems

SCHOOL OF BIOMETRICS Invited Speaker
Ursula Gather
Department of Statistics, University of Dortmund, Germany

We discuss filtering procedures for robust extraction of a signal from noisy time series as they occur in monitoring of vital parameters in intensive care. Moving averages and running medians are standard approaches to this, but they have shortcomings when large spikes (outliers) respectively trends occur. Modified trimmed means and linear median hybrid filters have been designed to combine advantages of both approaches, but they do not completely overcome the difficulties. Improvements can be achieved by methods of robust regression, which are applicable even in real time because of increased computational power and faster algorithms. Reviewing and extending recent work we present methods for robust online signal extraction and discuss their merits for preserving trends, abrupt shifts and extremes and for the removal of spikes.

References:

Bernholt, T., and Fried, R. (2003).
Computing the update of the repeated median regression line in linear time. Information Processing Letters 88, 111–117.

Davies, P. L., Fried, R., and Gather, U. (2004).
Robust signal extraction for on-line monitoring data. J. Statistical Planning and Inference, to appear.

Fried, R., Bernholt, T., and Gather, U. (2004).
Repeated median and hybrid filters. Technical Report, SFB475, University of Dortmund, Germany.

Gather, U., and Fried, R. (2004).
Robust estimation of scale for local linear temporal trends. PROBASTAT 2002, to appear.

How to Apply for the BIOSTAT 2006

The ITI Conference participants who wish to join the Meeting or School should register using the ITI Registration Form (http://iti.srce.hr/itireg.html). Please indicate on the Form your intention to participate in the School of Biometrics, in order to receive the School handouts at the Conference. For the ITI participants, there is no additional fee for the School. Submitted papers or poster abstracts (http://iti.srce.hr/submit.html) should reach ITI Conference secretariat within ITI deadlines and should follow ITI Instructions to authors (http://iti.srce.hr/itinstr.html). Accepted papers will be presented within the ITI topic: Data Mining, Statistics and Biometrics. The detailed program will be announced later, on the ITI web site at http://iti.srce.hr/.