Researchers at HUN-REN SZTAKI and Semmelweis University have developed a camera system driven by artificial intelligence that can continuously observe the posture and movements of preterm infants inside incubators without physical sensors or wired attachments. The institutions announced the development in a HUN-REN press release.
Why it matters
Tracking infants' movements and sleep positions in perinatal intensive care units is essential for assessing nervous system development and for early detection of potential neurological abnormalities. Current hospital practice largely relies on periodic, personal observations by neonatal nurses, which are intermittent and can be subjective.
The research team argues that a continuous, objective monitoring system can reduce observation gaps, provide more precise developmental data, and allow care and examinations to be timed better according to each infant’s natural sleep–wake cycles.
How the system works
The system analyzes video of newborns in incubators in real time and determines their posture and movements with centimetre-level accuracy. The software is trained to distinguish normal, developmentally appropriate movements from reactions that may indicate stress or other abnormalities. This capability enables clinicians to align necessary medical checks and interventions with times when infants are least disturbed.
The development was led by Földesy Péter in collaboration with the Neonatology Department of Semmelweis University.
Training data and methodology
Deep learning algorithms underpin the technology and were developed using an extensive video database. During the research, the team analyzed approximately 8,400 hours of video footage collected from 88 preterm infants to train the models to recognize movement patterns and possible deviations.
Publication
The results of the development and study were published in the journal Pediatric Research, which is part of the Nature family of journals.
Expected impact
Continuous, non‑contact, personalized monitoring could support gentler caregiving, promote faster development in preterm infants, and potentially shorten hospital stays by enabling more timely interventions. The press release emphasizes the technological capabilities and research findings; further clinical validation may be required before routine, wide‑scale implementation.
