Video-Based versus On-Site Neonatal Pain Assessment in Neonatal Intensive Care Units: The Impact of Video-Based Neonatal Pain Assessment in Real-World Scenario on Pain Diagnosis and Its Artificial Intelligence Application

Background: Neonatal pain assessment (NPA) represents a huge global problem of essential importance, as timely and accurate neonatal is indispensable for implementing management. Purpose: To investigate the consistency scores derived through video-based NPA (VB-NPA) on-site (OS-NPA), providing scientific foundation feasibility adopting VB-NPA results in real-world scenario gold standard clinical studies labels artificial intelligence (AI)-based (AI-NPA) applications. Setting: A total 598 neonates were recruited from pediatric hospital China. Methods: This observational study recorded who underwent one 10 painful procedures, including arterial blood sampling, heel fingertip intravenous injection, subcutaneous peripheral cannulation, nasopharyngeal suctioning, retention enema, adhesive removal, wound dressing. Two experienced nurses performed OS-NPA at 10-day interval double-blind scoring using Infant Pain Scale to evaluate level neonates. Intra-rater inter-rater reliability calculated analyzed, paired samples t-test was used explore bias assessors’ VB-NPA. The impact different label sources evaluated three state-of-the-art AI methods trained with given by VB-NPA, respectively. Results: intra-rater same assessor 0.976–0.983 across times, measured intraclass correlation coefficient. 0.983 single measures 0.992 average measures. No significant differences observed between an independent assessor. only caused limited accuracy loss 0.022–0.044 methods. Conclusion: effective way assess due its high compared could be labeling large-scale video databases training.