Abstract:

Acute hypotension is one of the complications which endanger patients’ lives in ICU. Earlier prediction for acute hypotension can help doctors find better medical treatment options. We propose a prediction model which bases on trendcomponent based Gaussian function fitting. We first use wavelet multiscale analysis to extract the trend components of the signals, whose function is fitted based on the Gaussian regression model. The Gaussian regression model is data driven. Its coefficients are used to describe the relationship between data, which are classified by the support vector machines (SVMs), and the function parameters are used as feature values. Experiments on a large dataset of patients prove that the new algorithm has better prediction results.

Key words: wavelet multiscale analysis;Gaussian process regression;function fitting;data driven