Deep learning using pose artificial intelligence (AI) may revolutionize neuro-telemetry in neonatal intensive care units (NICUs), according to a recent study published online in eClinicalMedicine.
Researchers from the Albert Einstein College of Medicine evaluated an AI-based computer vision system designed to track infant movements and predict neurological changes. The study analyzed video data linked to electroencephalograms from 115 infants across four urban NICUs between February 2021 and December 2022.
The findings revealed that the AI accurately predicted infant pose with high receiver operating characteristic area under the curves (ROC-AUCs) of 0.94 for cross-validation, 0.83 for held-out frames, and 0.89 for held-out infants. Additionally, the system effectively predicted sedation levels and cerebral dysfunction.
Dr. Felix Richter, the senior author, stated, "We envision a future system where cameras continuously monitor infants in the NICU, with AI providing a neuro-telemetry strip similar to heart rate or respiratory monitoring, alerting clinicians to changes in sedation levels or cerebral dysfunction." This innovation could enhance bedside care by allowing clinicians to review AI-generated insights alongside video footage.
Several authors of the study are listed as inventors of Pose AI, which aims to monitor neonatal mental status and has filed a provisional patent application.