AI Uncovers Serious Neurologic Changes in NICU Babies Using Video Data: A Breakthrough in Healthcare Technology

Jakarta, INTI - A team of medical experts, scientists, and engineers at the Mount Sinai Health System has successfully trained an AI-based pose-recognition algorithm to monitor babies' movements in neonatal intensive care units (NICUs), identifying neurologic changes in infants with high accuracy using only video data. Published on November 11 in The Lancet's eClinicalMedicine, this groundbreaking AI tool could introduce a minimally invasive, continuous neurologic monitoring method in NICUs, providing critical real-time insights for clinicians to better assess infant health.

Each year, over 300,000 newborns are admitted to NICUs across the United States, where they face high risks for neurological decline. These declines can occur suddenly and cause serious outcomes, but continuous neurologic monitoring technology remains limited in most NICUs. While cardiorespiratory functions like heart and lung activity are constantly tracked, neurologic status is often evaluated through intermittent physical exams, which can miss sudden changes and are not always precise.

To address this gap, the Mount Sinai team developed an approach using a method called "Pose AI," in which the algorithm interprets anatomical landmarks to track an infant’s movements from video data. This method has already proven effective in various fields, such as sports and robotics, and is now adapted for tracking neurologic health in NICU infants.

AI "Pose" Tracks Baby Movements with High Accuracy

The team trained Pose AI using over 16.9 million seconds of video footage from 115 infants in the NICU at Mount Sinai Hospital. Throughout training, the algorithm was supplemented with EEG (electroencephalography) data as a benchmark to predict two critical neurologic conditions: sedation and brain dysfunction. Through video data analysis, Pose AI accurately detected neurologic indicators associated with both conditions.

Dr. Felix Richter, a pediatrician and researcher at Mount Sinai's Department of Pediatrics, explained, "Our study shows that applying an AI algorithm to NICU cameras can detect neurologic changes in infants more quickly. This technology could enable faster interventions, which has the potential to improve clinical outcomes for patients."

Pose AI Performs Well Across Various Lighting Conditions

One impressive feature of Pose AI is its robustness across different lighting conditions, including nighttime and for babies undergoing phototherapy. Additionally, the algorithm showed a significant correlation between infant movement indexes and both gestational and postnatal age, making it an age-sensitive assessment tool crucial for neurologic evaluation in infants.

"This approach does not replace essential doctor or nurse assessments but provides reliable, continuous data in a clinical context," Dr. Richter emphasized. Pose AI offers neurologic indicators that can be tracked similarly to a neuro-telemetry strip, like heart rate or respiratory monitoring in NICUs. This allows doctors to review video footage and receive AI-generated insights on sudden neurologic changes when necessary.

Challenges in AI Implementation in NICUs

Despite promising results, the study has some limitations. The Pose AI algorithm was developed using data from a single hospital, Mount Sinai Hospital, so further testing across different NICUs is needed to ensure optimal performance with diverse video and camera setups.

In addition, the research team plans to conduct clinical trials to measure the impact of AI-assisted monitoring on NICU care quality. They also intend to adapt this technology to monitor neurological conditions in other patient groups, including adults, and expand it to detect a wider range of neurologic issues.

AI Technology's Impact on NICU Infant Care

Mount Sinai's approach marks a significant milestone in applying AI to support neonatal healthcare. Pose AI offers a comprehensive, real-time neurologic monitoring solution, allowing early detection of potentially life-threatening neurologic changes. By reducing reliance on intermittent physical exams, this technology could increase both the speed and accuracy of medical interventions.

Dr. Girish N. Nadkarni, Chief of Data Driven and Digital Medicine at Mount Sinai, noted, “AI tools have accelerated clinical advancements across the Mount Sinai Health System, such as reducing hospital stays, aiding in cancer diagnostics, and providing real-time care based on physiological data from wearables. We’re excited to bring this safe, effective, non-invasive AI tool into the NICU to improve outcomes for our smallest, most vulnerable patients.”

Future Potential and Benefits of Pose AI Technology

Pose AI’s ability to detect early neurologic issues in infants has the potential to enhance NICU safety and overall care quality. The data generated by Pose AI can provide critical indicators that help doctors make timely and accurate decisions in infant care.

Looking ahead, the development team aims to integrate Pose AI with other monitoring systems, such as surveillance cameras and wearable devices, to collect and interpret more comprehensive data for medical teams. Continuous neuro-telemetry provided by Pose AI is anticipated to become an essential tool in treating infants with complex health conditions in NICUs.

With rapid technological progress, it is hoped that AI systems like Pose AI will gain widespread adoption across healthcare institutions, significantly benefiting NICU infants. This technology also opens the door for neurologic monitoring applications in adult patient care, improving healthcare quality and efficiency across the board.

Source: ScienceDaily