Determining the Health Condition of Newborns Using Apgar Score and Thermal Camera Imaging

APGAR (Appearance, Pulse, Grimace, Activity, Respiration) Score is a system that evaluates the post-birth health status of newborns and is scored between 0-10. This score, which consists of five parameters: appearance, pulse, grimace, activity and respiration, is obtained by scoring each parameter between 0-2. The condition of newborns with an APGAR score of 7-10 is considered reassuring, a score of 4-6 is considered moderately abnormal and a score of 0-3 is considered low and critical. This study aims to automatically determine the APGAR score of newborn babies with thermal camera imaging and artificial intelligence support and to classify the newborn according to the APGAR score. For this purpose, data was collected from a total of 22 newborns whose gestational age was between 28-42 weeks. Thermal images of newborns were preprocessed in the MATLAB software environment and first the babies were separated from the background. After this stage, important features determined as a result of the literature research were extracted. The extracted features and the APGAR scores obtained from nurses, which are accepted as gold standards, were used in the Python software environment for classification. Stacking Classifier models were used in the regression and classification stages. Among the classification algorithms, the most successful model in distinguishing risky and healthy babies was the Stacking Classifier with 88.59% (+/- 0.0264) accuracy. The proposed method may assess APGAR score non-invasively, non-contact, objectively and early detection of risky newborns was provided.