Mammoths: A Vital Nutrient for Early American Societies

Scientists have uncovered the first direct evidence that ancient Americans relied primarily on mammoth and other large animals for food. Their research sheds new light on both the rapid expansion of humans throughout the Americas and the extinction of large ice age mammals. Scientists have uncovered the first direct evidence that ancient Americans relied primarily
HomeHealthRevolutionary AI Technology Identifies Neurological Changes in NICU Babies Through Video Analysis

Revolutionary AI Technology Identifies Neurological Changes in NICU Babies Through Video Analysis

Study findings could pave the way for wider neuro-monitoring technologies in intensive care units worldwide.

A collaborative team of doctors, researchers, and engineers at Mount Sinai has developed a deep learning pose-recognition algorithm that analyzes video footage of infants in the neonatal intensive care unit (NICU) to accurately monitor their movements and track important neurological indicators.

Results from this innovative artificial intelligence (AI) tool, published on November 11 in Lancet’s eClinicalMedicine, may facilitate a minimally invasive and scalable approach to ongoing neurological monitoring in NICUs, offering critical real-time insights into infant health that were previously unattainable.

In the United States, over 300,000 newborns are admitted to NICUs each year. Infant alertness is regarded as a crucial component of the neurological evaluation, indicating the health of the central nervous system. Unfortunately, neurological decline in NICUs can occur without warning, leading to severe consequences. While cardiorespiratory telemetry continuously tracks the heart and lung functions of infants in the NICU, neurotelemetry has been challenging to implement in most NICUs, despite years of advancements in electroencephalography (EEG) and specialized neuro-NICUs. Neurological assessments are performed sporadically through physical examinations that may overlook subtle changes.

The team at Mount Sinai theorized that a computer vision approach to monitor infant movements could predict neurological alterations in NICUs. “Pose AI” is a machine learning technique that identifies anatomical markers from video data, significantly advancing fields like athletics and robotics.

They trained an AI algorithm using over 16,938,000 seconds of video from a varied group of 115 NICU infants receiving continuous video EEG monitoring at The Mount Sinai Hospital. Their findings revealed that Pose AI could reliably track infant anatomical landmarks from the video footage. Additionally, they utilized these landmarks to precisely predict two critical conditions—sedation levels and cerebral dysfunction.

“Even though numerous neonatal intensive care units have video cameras, they have not yet leveraged deep learning for patient monitoring,” stated Felix Richter, MD, PhD, lead author and Instructor of Newborn Medicine in the Department of Pediatrics at Mount Sinai. “This study illustrates that deploying an AI algorithm with cameras monitoring infants continuously in the NICU effectively identifies neurological changes early, potentially leading to quicker interventions and improved patient outcomes.”

The research team was impressed with how effectively Pose AI performed under varying lighting conditions (daytime, nighttime, and during phototherapy) and from diverse angles. They were also surprised by the connection between their Pose AI movement index and both gestational and postnatal ages.

“It is crucial to understand that this method does not substitute for the assessments made by physicians and nurses, which are essential in the NICU. Instead, it enhances these evaluations by providing continuous data that can be acted upon in specific clinical situations,” Dr. Richter explained. “We foresee a future system in which cameras continuously observe infants in the NICU, with AI offering a neuro-telemetry strip akin to monitoring heart rate or respiratory functions, complete with alerts for changes in sedation or cerebral dysfunction. Clinicians could review video feeds and insights generated by AI as necessary, providing a straightforward and interpretable tool for bedside care.”

The team acknowledged the study’s limitations, including the fact that AI models were trained on data from a single site, indicating that this algorithm and its neurological predictions require further validation using video data from other institutions and camera systems. They plan to test this technology in additional NICUs and to conduct clinical trials assessing its impact on patient care. Furthermore, they are exploring its potential application to other neurological disorders and its expansion to adult patient care.

“At Mount Sinai, we are dedicated to exploring the potential of artificial intelligence to enhance patient care,” said Girish N. Nadkarni MD, MPH, System Chief of Data Driven and Digital Medicine, Director of the Mount Sinai Clinical Intelligence Center, and Director of The Charles Bronfman Institute for Personalized Medicine, who is also a co-author of the study. “AI tools are already transforming clinical practices within the Mount Sinai Health System, for instance, by reducing hospital stays, minimizing readmissions, assisting in cancer diagnosis and treatment, and providing real-time patient care based on physiological data from wearables, among other applications. We are thrilled to introduce this safe, non-invasive, and effective AI tool to the NICU, enhancing outcomes for our smallest and most vulnerable patients.”