Date of Award
Master of Science in Statistics
Computer Science and Statistics
The objective of this thesis is to explore the relationship between cry acoustic data and neurological development. The methods used in this project were general linear models as well as XGBoost models. Additional methods explored include general linear mixed effects models and combining mixed effects estimation with machine learning algorithms. A description and simulation to compare those methods is provided, and a discussion on when using mixed effects with machine learning could be beneficial. An analysis was carried out to predict autism spectrum disorder using cry recording data from newborn infants, as well as a simulation study to understand how sample size can impact the results of that modeling. Another analysis was carried out to predict 12 month development index from the cry acoustic data. Cry acoustic data seems to decrease out of sample variation when predicting ASD, but not clear advantage is seen when predicting mental development index.
Prows, Broderick, "PREDICTING NEUROLOGICAL DEVELOPMENT OUTCOMES WITH CRY ACOUSTIC DATA" (2022). Open Access Master's Theses. Paper 2156.
Available for download on Friday, May 17, 2024