Neonatal Bowel Sound Detection Using Convolutional Neural Network and Laplace Hidden Semi-Markov Model

Abdominal auscultation is a convenient, safe and inexpensive method to assess bowel conditions, which essential in neonatal care. It helps early detection of dysfunctions allows timely intervention. This paper presents sound assist the auscultation. Specifically, Convolutional Neural Network (CNN) proposed classify peristalsis non-peristalsis sounds. The classification then optimized using Laplace Hidden Semi-Markov Model (HSMM). validated on abdominal sounds from 49 newborn infants admitted our tertiary Neonatal Intensive Care Unit (NICU). results show that can effectively detect with accuracy area under curve (AUC) score being 89.81% 83.96% respectively, outperforming 13 baseline methods. Furthermore, HSMM refinement strategy proven capable enhance other models. outcomes this work have potential facilitate future telehealth applications for source code be found at: https://bitbucket.org/chirudeakin/neonatal-bowel-sound-classification/