Artificial Neural Network for Identification of Infant Feeding Tracking Using the Smart Bottle System

In this work, we present the results of a comparison of simple artificial neural network (FFNN) designs intended to identify infant bottle-feeding events and appropriate feeding volume recording intervals using accelerometer data recorded from a custom designed "Smart Bottle" system. To properly identify and distinguish these events with an accuracy of 99.8%, while accommodating the constraints of the deployment environment, two concurrent FFNNs were implemented.