Peak detection of the plasma cortisol time series by syntactic pattern recognition

A syntactic peak recognition algorithm was applied to identify the significant peaks in the sampled profiles of cortisol concentration for 29 normal subjects and 49 patients with Cushing´s syndrome. To estimate the parameters of the criterion for a significant peak, the random measurement error in the laboratory assay must first be established, and the first derivative of the cortisol time series must be calculated. This process is described. The syntatic peak recognition algorithm was applied to the time series to identify significant secretory peaks. The mean of the number of peaks in the normal group proved to be significantly different from the respective means of the number of peaks in the individual patient subgroups. This result may be useful in the diagnosis of Cushing´s syndrome