Research

Medical Sciences

Title :	Development of AI based neonate-infant cry analyzer and assessment of link between protophones and maternal sleep during pregnancy
Area of Research :	Medical Sciences
Focus Area :	AI in Clinical Diagnostics
Principal Investigator :	Dr. Kamalesh K Gulia, Sree Chitra Tirunal Institute for Medical Sciences and Technology, Thiruvananthapuram, Kerala (695012)
Co-PI:	Dr. Arun Anirudhan, Sree Chitra Tirunal Institute for Medical Sciences and Technology, Thiruvananthapuram, Kerala (695012), dr. Soumya Sundaram, Sree Chitra Tirunal Institute for Medical Sciences and Technology, Thiruvananthapuram, Kerala (695011)
Contact info :	kkgulia@sctimst.ac.in; arunanirudhan@sctimst.ac.in; drsoumya@sctimst.ac.in
Timeline Start Year :	2023
Timeline End Year :	2026
Total Budget (INR):	44,06,875

Details

Executive Summary :

Cry is a crucial acoustic signal produced by newborns, providing hidden information about their state of wellbeing. It is difficult to decode the underlying message in cries, as language is not developed at birth and babies are born with mechanisms of cry to convey discomfort and call for attention. The vocalization pattern in cry varies but it is hard to differentiate cry types. Neonates sleep for about 80-85 minutes in bouts of 2-3 hours, which gradually reduces and infants begin to spend more time staying awake during infancy. Pediatricians follow the Wessel rule of 3 for colic-related excessive cries, which states inconsolable crying for 3 hours per day, at least 3 times a week for at least 3 weeks. However, most infants do not comply with this rule. The Mel Frequency Cepstral Coefficient (MFCC) algorithm is useful in classifying cry signals in machine learning (ML), while advanced Artificial Intelligence tools like deep learning will be useful in sound detection from complex spectrograms. A proposed AI deep learning tool of convolutional neural network (CNN) is proposed to generate suitable algorithms to classify cry types. The researchers propose to derive a large national CRY database repository for training the AI algorithm for identification of normal vs abnormal cries and develop a mobile-based CRY app to track the wellbeing of infants. Emerging concepts on protophones suggest that these precursors of human speech are also observed during early infancy along with cries. Derive ontogenetic profiling of protophones in infants with normal and deviant cries with profiling of maternal sleep during pregnancy. This study aims to develop a non-invasive real-time CRY analyzer app that can help parents and caregivers understand their child's wellbeing remotely. By utilizing AI tools and generating a large-scale dataset, this project contributes to the field of CRY signal analysis and prediction.

Equipments :

Microphones with sound interface-20 ( Rs.942555.00 ) , Microphones with sound interface-20 ( Rs.0.00 ) , Microphones with sound interface-20 ( Rs.0.00 )

Organizations involved

Implementing Agency :	Sree Chitra Tirunal Institute for Medical Sciences and Technology, Thiruvananthapuram, Kerala
Funding Agency :	Department of Science and Technology (DST)
Source:	Department of Science and Technology (DST)