Computer-based statistical, pattern recognition of Doppler spectral waveform kinematics during fetal respiration

A computer-based statistical pattern recognition program from digitized spectral waveform envelopes of amniotic fluid displacement during fetal 'respiration'' was developed to extract discriminant features of respiratory maturation. The accuracy of the model reached 98% in estimating maturity in 37 healthy controls. A subset of high-risk subjects was classified with 100% accuracy in relation to maturity and growth. The incorporation of computer algorithms to extract significant kinematic variables (p < ,05) using pattern-based recognition of spectral signal features may provide a method to classify respiratory-related maturation in medical practice and obtain objective clinical indices of abnormal pulmonary states