Uma ferramenta computacional para auxílio didático e avaliação metodológica de profissionais na prática clínica em aleitamento materno

Abstract

Breastfeeding is essential for children’s health and survival, as it provides a safe, natural, nutritious, and sustainable source of nourishment. Breast milk contains antibodies that protect against various common childhood diseases, such as diarrhea and respiratory infections. Inadequate breastfeeding is associated with 16% of annual infant deaths, and breastfed children tend to perform better on intelligence tests while having a lower likelihood of developing obesity or being overweight later in life. Additionally, women who breastfeed have a reduced risk of developing cancer and type II diabetes. However, accurately measuring the amount of breast milk ingested by a baby remains a challenge. Therefore, there is a clear need for an instrument that not only evaluates specific protocols but also offers educational support to breastfeeding professionals, promoting more precise documentation and more effective management. The effectiveness of breastfeeding management can be limited by reliance on clinical observation and professional experience, highlighting the need for more accurate tools. This study proposes a prototype of a computational solution that estimates the amount of breast milk ingested based on the sound of the baby’s swallowing, providing a quantitative analysis that improves breastfeeding monitoring. Consequently, this solution serves as an educational resource, offering feedback to breastfeeding profes ionals, allowing for the reevaluation of practices and protocols. The development of this prototype involved applying knowledge of breastfeeding protocols, pattern recognition, and physical computing. The prototype of the educational product developed in this master’s thesis was evaluated based on the performance of the Artificial Intelligence model used, with error measured through specific metrics. The results indicated that the Artificial Intelligence model of the educational product exhibits error rates within acceptable limits.