Infant Health Risk Detection Using Deep Learning-Based Cry Analysis

Infant crying is a primary means of communication, but distinguishing between normal and abnormal cries is challenging for caregivers and healthcare providers. This study proposes an automated system for infant health risk detection using deep learning-based cry analysis. The system extracts acoustic features such as Mel Frequency Cepstral Coefficients (MFCC) and pitch from the cry signals of infants and it utilize the Convolutional Neural Networks (CNN) for classification. The results of experiment show that the CNN model has achieved higher accuracy compared to the old  machine learning methods like K-Nearest Neighbors (KNN). The proposed approach effectively classifies infant cries into normal and abnormal categories, enabling early detection of potential health issues.