Investigation of Relation Between Hypoxic-Ischemic Encephalopathy and Spectral Features of Infant Cry Audio

• Published in: Journal of voice Vol. 38; no. 6; pp. 1288 - 1295
• Main Authors: Satar, Mehmet, Cengizler, Caglar, Hamitoglu, Serif, Ozdemir, Mustafa
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.11.2024
• Subjects: Acoustics; Age Factors; Cry; Crying; Diagnosis; Female; Humans; Hypoxia-Ischemia, Brain - diagnosis; Hypoxia-Ischemia, Brain - physiopathology; Hypoxic-ischemic encephalopathy; Infant; Infant Behavior; Infant, Newborn; Machine Learning; Male; Newborn; Otolaryngology; Predictive Value of Tests; Signal Processing, Computer-Assisted; Sound Spectrography; Spectral; Time Factors; Spectral; Newborn; Cry; Diagnosis; Hypoxic-ischemic encephalopathy
• ISSN: 0892-1997; 1873-4588; 1873-4588
• DOI: 10.1016/j.jvoice.2022.05.015

Despite advances in medical technologies, Hypoxic-Ischemic Encephalopathy (HIE) continues to be a problem for neonatal intensive care units. Analysis of crying sounds may be a valuable tool for predicting neonatal disease. However, the characteristics of crying in newborns with HIE are still unclear. One of the factors limiting the ability to focus on that subject is the lack of commercially available infant cry database for research. Also, another reason that complicates the classification is the varying characteristics of infant cry. Accordingly, crying sounds were recorded from 35 infants and demographic characteristics of the study groups are presented as well as the numerical representation of spectral features. Experiments reveal that the existence of HIE causes distinctive variation in energy, energy entropy and spectral centroid features of the utterances; which leads us to conclude that the presented combination of spectral features would function well with any supervised or unsupervised machine learning algorithm.