[Data reduction, smoothing and base line correction preceding an automatic evaluation of ECG signals (author's transl)].

In storing electronically digital data of signals, too many items are used. Especially the great number of data produced in the analogue-digital-conversion of the ECG leads to serious problems in long distance transmission and storage. It is proposed to described such ECG signals through polynominal fragments, to which such a position across the noise-level distorted data is given that at the same time the data are levelled in a flexible manner. Based on these calculated reduced data, shape recognition problems can be solved more easily. As an example, a suggestion is given to construct the isoelectric segment of an ECG.