UC San Diego

Breastfeeding is a natural biologic function that benefits both mothers and infants by protecting their health and development. Since the 1950s, breastfeeding rates have dropped dramatically, despite nearly 80\% of mothers attempting to breastfeed. Breastfeeding cessation is caused by many factors from both mother and infant, with many reporting nipple pain, poor milk transfer, and poor infant weight gain as a few of the many contributors. An infant's ability to suckle in a coordinated manner is a key element to successful breastfeeding. When uncoordinated or irregular, an infant's suckling may cause pain, poor latch, and poor milk transfer ultimately increases a mother's risk of breastfeeding cessation by disrupting her ability to nurse or pump. In the past two decades, many devices and systems have been developed to address the issue of infant oral motor coordination, particularly in pre-term infants that lack of the necessary oral motor developments due to their premature birth. Abnormal suckling behavior in full term infants have been largely overlooked as many have turned to surgical intervention to resolve congenital oral dysfunction, with little evidence of long-term benefits. Despite rising trends in the last decade in surgical interventions to resolve breastfeeding issues, breastfeeding cessation rates continue to climb. Lack of standardized objective measurement tools for general screening of infant suckling to guide data-driven intervention remains a challenge within the clinical community.This dissertation studies non-nutritive suckling behavior in full-term healthy infants. To address the need for standardize objective measurement tools, a non-nutritive suckling measurement system was designed and developed to enable real-time measurement of infant suckling vacuum. An accompanying software was created to enable clinicians to interact and interface with the data in real-time for rapid diagnosis and analysis in regions of interest. The system was used in clinical evaluation of 91 healthy full term infants to establish normative data for non-nutritive suckling. Once normative data was sufficiently collected, data from abnormal suckling behavior caused by a common congenital condition were studied and analyzed. Extensive signal processing was performed to extract characteristic features from non-nutritive suckling signals such as max vacuum, mean vacuum, suckling frequency, burst duration, sucks per burst, and three principal frequency components describing signal shape. Machine learning algorithms were used to assist with anomaly detection to determine of abnormal suckling behavior can be automatically determined based on normative data. Case evaluations are studied in conjunction with clinical notes and assessments to determine congruence or disconsensus between traditional examinations and objective measurements.Confounding evidence of clinical inconsistency using standard evaluation methods are discussed as apart of the larger goal of shedding light on the degree of subjectivity that affects intervention and diagnosis of breastfeeding difficulties caused by infant suckling irregularities. Finally, the work is summarized and future directions are described to lay the foundation for continued advancement in the field, technology, and clinical practices.