Analysis of dispersion and principal component analysis of babblings’ signals from moderate preterm and term infants

Abstract

There are different milestones expected to achieve in each month while infants are growing. During the first year of live they produce vocal sounds known as crying and babbling, which requires the coordination and development of different systems. These productions are being studied to find patterns which can be used to identify infants with developmental impairments or with normal development. Previous studies focused on crying signals, even though, none of the methodologies used have been established as a reliable method for clinical diagnosis of developmental impairments. This paper proposes a statistical analysis of babblings’ signals in order to identify patterns to differentiate between infants’ patients and controls. The size of the sample was 180 signals, 55 from 8 moderate preterm infants, and 125 from 22 term infants. A short-time Fourier transform was applied to each signal and a statistical validation was made using analysis of dispersion and principal component analysis. The results indicate ranges with significant difference between patients and controls sets.