Internet of medical things for abnormality detection in infants using mobile phone app with cry signal analysis

Detecting the baby’s cry sounds is significant and is the first step that enables effective diagnosis in the branch of pediatrics. Despite the complexity in the analysis of the baby’s cry signal, an automated cry signal segmentation system can be introduced for the diagnosis of earache, colic pain, cold, diaper rashes, or due to hunger. This is a challenging task as this type of automated cry sound segmentation algorithm is dependent on the wavelet coefficients extracted from the cry signal. These coefficients are the inputs to train the cry signal-oriented diagnostic system. A completely computerized segmentation algorithm is designed to extract the details and approximation coefficients of the cry signal during the expiration and inspiration process. These coefficients are used to train the convolutional neural networks (CNN). The prime focus of this work is to devise a smartphone-based app that will record the baby’s cry signal, segment it using the wavelet transform, and classify them using CNN based on the diagnosis made to identify the earache, colic pain, cold, diaper rashes, fever, respiratory problem or hunger. This indigenous smartphone app will enable the young mothers to identify the problem existing with their infants and facilitate an easy nurturing of the newborn. This non-contact type of diagnosis finds a lot of importance in the present scenario, where the COVID-19 social distancing is followed enabling the physician, infant, and mother to be devoid of the fear of this pandemic situation. The main objective of this proposal is to design a cry signal based infant diagnostic system which focuses on scrutinizing the neonatal pathologies by extracting the features present in the signal of the baby’s cry in a realistic clinical environment. This mobile app once developed, will be a part of the internet of medical things