Unsupervised hierarchical fuzzy clustering methods in forecasting medical events from biomedical signals

Many problems in the field of biomedical signal processing can be reduced to a task of state recognition and event forecasting. Examples can be found in tachicardia detection from ECG signals, epileptic seizure or psychotic attack prediction from an EEG signal, and prediction of vehicle drivers falling asleep from both signals. The problem generally treats a set of ordered measurements and asks for the recognition of some patterns of observed elements which will forecast an event or a transition between two different states of the biological system. In this paper we propose to apply clustering methods to grouping discontinuous related temporal patterns of a continuously sampled measurement. The vague switches from one stationary state to another are naturally treated by means of fuzzy clustering. In such cases an adaptive selection of the number of clusters (the number of underlying semi-stationary processes in the signal) can overcome the general non-stationary nature of biomedical signals and enables the formation of a warning cluster. The algorithm suggested for the clustering is a new recursive algorithm for hierarchical-fuzzy partition. The algorithm benefits from the advantages of hierarchical clustering while obtaining fuzzy clustering rules. Each pattern can have a non-zero membership in more than one sub-data-sets in the hierarchy. Optimal feature extraction and reduction is reapplied for each sub-data-set. A `natural' and feasible solution to the duster validity problem is suggested by combining hierarchical and fuzzy concepts. The algorithm is shown to be effective for a variety of data sets with a wide dynamic range of both covariance matrices and number of members in each class. The new method is applied to the forecasting of biomedical events like generalized epileptic seizures from the EEG and heart rate signals.