Traditional neonatal monitoring relies on contact-based sensors like ECG electrodes and pulse oximeters, which pose risks such as skin irritation, motion artifacts, and infection exposure. Non-contact methods—computer vision, infrared thermography, and near-field sensing—offer a promising alternative but suffer from limited accuracy, motion sensitivity, and suboptimal spatiotemporal coverage in NICUs.

 

We are developing an AI-driven, multi-modal non-contact neonatal health monitoring system that integrates seamlessly into NICU radiant warmers. Our system utilizes:

• Computer Vision & Deep Learning: AI-based object identification and neonatal skin segmentation ensure precise vitals extraction.

• Infrared Thermography: Real-time, contact-free temperature tracking with high sensitivity.

• Near-Field & Optical Sensing: Continuous respiratory rate and cardiac signal detection, mitigating motion-induced artifacts.

• Sensor Fusion & AI Models: A robust algorithm pipeline combines multi-modal inputs to enhance reliability and accuracy.

 

Our AI models for neonatal motion detection, vitals tracking, and behavioral state identification have been successfully validated in initial trials on 40 neonates. Next steps involve clinical validation with advanced technology prototypes, ensuring scalability and real-world efficacy in neonatal intensive care.

​

Clinical Partner: Saveetha Medical College, Chennai