Real-time analysis of foetal heart rate patterns using a computer system

Abstract

The foetal heart rate and uterine contraction curves provide both continuous and real-time information on the foetus. In recent years, attempts have been made to automatically diagnose foetal distress using a computer to provide direct analysis of this information. In the present study, we performed pattern-analysis of the foetal heart rate and uterine contraction curves using a minicomputer on 476 mothers who gave birth at the hospital attached to Hamamatsu University School of Medicine from April 1978 to March 1980. We developed the algorithm for these curves. The algorithm was roughly divided into two groups, original signal process and pattern-recognition algorithms. Following the autocorrelation and noise reduction of original signals, the algorithm was analysed. The algorithmic analysis consisted of the calculation of baseline f.h.r. recognition of deceleration, recognition of uterine contraction, and classification of the deceleration pattern. By using these algorithms, the true diagnostic distribution rate was obtained: early deceleration: 76%, variable deceleration: 88%, late deceleration: 72%, and prolonged deceleration: 81%.