Neonatal Neurological Observation With Video AI

Overview

NeoNOVA is a multi-site, prospective, single-arm, silent observational study to determine: among (Population) infants admitted to newborn services during their inpatient hospital stay, whether (Intervention) continuous bedside non-contact high definition video running real-time AI analysis of anatomic landmarks and movement, (Comparison) compared against human-labeled video frames and standardized clinical exams, will (Outcome) accurately localize infant anatomic landmarks (primary objective; outcome median position error in pixels) and demonstrate a statistically significant association between a video-derived movement index and clinical measures of patient neurological exams (secondary objective; outcomes N-PASS and modified Sarnat exams).

Detailed description

To fill this critical gap in neonatal care, the investigators developed and validated NeoPose, a low-cost, non-invasive, computer vision digital health tool to continuously monitor infants using real time video streams. NeoPose uses Pose Artificial Intelligence (AI) for an explainable approach to measure, quantify, and analyze infant movement. From the vectorized movement, investigators can accurately confirm the presence of encephalopathy and quantify the degree of sedation. The explainable AI platform enables continuous neuromonitoring with AI-driven alerts, suspicious event replay, movement comparisons, and training on a vast dataset of normal and abnormal infant movements far beyond what any provider could witness.

The Neonatal Neurological Observation with Video AI (NeoNOVA) study is a multi-site, prospective, single-arm, pragmatic, silent observational study to evaluate the performance of NeoPose and AI-derived insights in real world settings. NeoNOVA will deploy a bedside video monitoring system (ArtemisAI Platform) that continuously, passively video records the subject from enrollment to discharge. The study will prospectively validate the AI system's tracking accuracy against ground-truth human-labeled video frames (primary objective; outcome median position error in pixels), will evaluate the association between a video-derived movement index and standardized bedside assessments of encephalopathy, pain, and sedation (secondary objective; outcomes N-PASS and modified Sarnat scales), and will support hypothesis-generating research on novel video prediction algorithms for outcomes like sepsis and need for respiratory support (tertiary objective). The study operates in "silent mode," where AI outputs are not shown to the patient's clinical team. Findings are intended to support a structured clinical evidence generation plan for a Software as a Medical Device (SaMD) designed for continuous, non-contact neurological monitoring in the NICU.

Interventions

• Device Continuous bedside video monitoring with AI anatomic landmark tracking for neurologic monitoring
  A non-contact, passive bedside video recording system is mounted adjacent to the infant's crib or incubator. The device continuously captures video data from enrollment to hospital discharge or withdrawal. The device runs AI models to track infant anatomic landmarks and calculate a continuous movement index. The trial runs in "silent mode," where AI outputs are not shown to the patient's clinical team and do not influence care.

Primary outcome measures

• AI Anatomic Landmark Tracking Accuracy [Time frame: At study completion, an average of 1 week.]

Secondary outcome measures (7)

• Movement Index - Encephalopathy measured by modified Sarnat exam [Time frame: Through study completion, an average of 1 week.]
• Movement Index - N-PASS [Time frame: Through study completion, an average of 1 week.]
• Movement Index - Sedative Exposure [Time frame: Through study completion, an average of 1 week.]
• Movement Index - Chronological Age at Video [Time frame: Through study completion, an average of 1 week.]
• Movement Index - Gestational age at birth [Time frame: Through study completion, an average of 1 week.]
• Movement Index - Sleep state [Time frame: Through study completion, an average of 1 week.]
• Movement Index - EEG evidence of cerebral dysfunction [Time frame: Through study completion, an average of 1 week.]

Eligibility criteria

Inclusion criteria

• Signed and dated informed consent from at least one parent or legally authorized representative (LAR) who is at least 18 years old.
• Parent/LAR expresses willingness to comply with study procedures for the duration of the infant's hospital stay.
• Infant of any sex (including intersex/undetermined) admitted to newborn services (including the NICU) at a participating hospital.

Exclusion criteria

• Parents or LAR unable to provide informed consent or are under the age of 18.
• Non-viable neonates

Criteria are shown verbatim from the registry (in English). Final eligibility is always assessed by the study center.

Healthy volunteers: Yes

Study design

Observational model

Cohort

Study locations

United States · 2 centers

• Mount Sinai Hospital — New York
• Weill Cornell Medicine / NewYork-Presbyterian Hospital — New York

Publications

• Feng R, Richter F, Mari E, Gleason A, Le C, Kellner CP, Shrivastava RK, Fields M, Rapoport BI, Bederson JB, Schadt EE, Glicksberg BS, Richter F, Dangayach NS. Artificial Intelligence Monitoring of Neurological Status From Patient Videos in the Neuroscience Intensive Care Unit. Neurosurgery. 2026 Jan 14. doi: 10.1227/neu.0000000000003899. Online ahead of print. PMID 41532764
• Gleason A, Richter F, Beller N, Arivazhagan N, Feng R, Holmes E, Glicksberg BS, Morton SU, La Vega-Talbott M, Fields M, Guttmann K, Nadkarni GN, Richter F. Detection of neurologic changes in critically ill infants using deep learning on video data: a retrospective single center cohort study. EClinicalMedicine. 2024 Nov 11;78:102919. doi: 10.1016/j.eclinm.2024.102919. eCollection 2024 Dec. PMID 39764545

Identifiers

NCT: NCT07628829 · STUDY-25-01036

Primary sources (government registries)

• ClinicalTrials.gov (US NIH)

View this study on ClinicalTrials.gov ↗