Methodology for validating blood pressure event detection algorithms

The methodology for a generating a waveform test set and producing a statistical representation of blood pressure event detection algorithm performance is demonstrated. A subset of the Massachusetts General Hospital (MGH) database files has been selected as an illustrative group of waveforms representative of a range of clinical and physiological conditions. Several dicrotic notch detection algorithms were applied to three different blood pressure waveform test sets and the accuracy of detection was determined. Comparison of the algorithms´ statistical performance for each test set led to the assembly of an appropriate grouping of physiological data for validation of blood pressure event detection techniques for a valid statistical representation of algorithm performance