Video-based detection of cessation of breathing in pre-term infants using machine learning

2026-07-06 · Source: Takara TLDR - Daily AI Papers · Field: Health & Wellbeing — Clinical Care & Medical Practice, Medical Specialties & Subspecialties, Medical Devices & Health Technology · Depth: Expert, quick

Summary

A study investigated non-contact video monitoring for detecting cessation of breathing (COBE) in pre-term infants, addressing challenges with traditional contact-based sensors in neonatal intensive care units (NICU). Researchers utilized video and clinical recordings from 30 pre-term infants, extracting respiratory motion from dynamically tracked torso regions to create camera-derived time-series signals. They trained "camera-only" models using Residual Network (ResNet) architectures, achieving a balanced accuracy of 76.9% for COBE detection. "Hybrid" models, which combined video-derived features with impedance pneumography (IP), ECG-derived respiration (EDR), and PPG-derived respiratory envelope, demonstrated enhanced performance. Specifically, combining video with IP boosted balanced accuracy to 90.6%, outperforming either modality alone. These findings confirm that video-derived signals contain clinically relevant respiratory features and significantly improve COBE detection when integrated with conventional physiological signals, supporting video as a robust complementary modality for automated neonatal respiratory monitoring.

Key takeaway

For AI Scientists and Research Scientists developing neonatal monitoring systems, this research indicates that integrating non-contact video with impedance pneumography significantly enhances apnoea detection. You should consider incorporating video-derived respiratory features into your machine learning models to improve robustness and overcome limitations of contact-based sensors. This approach offers a path to more reliable, less invasive monitoring in NICU environments.

Key insights

Non-contact video monitoring effectively detects breathing cessation in pre-term infants, especially when combined with impedance pneumography.

Principles

• Video provides complementary respiratory data.
• Hybrid monitoring improves detection accuracy.
• Non-contact methods reduce sensor issues.

Method

Respiratory motion was extracted from dynamically tracked torso regions in video. ResNet models were trained on these camera-derived time-series signals, alone or combined with physiological data.

In practice

• Integrate video with existing IP systems.
• Explore ResNet for motion-based vital sign detection.
• Develop non-contact NICU monitoring solutions.

Topics

• Neonatal Monitoring
• Apnoea Detection
• Machine Learning
• Video Analytics
• ResNet Architectures
• Impedance Pneumography

Best for: Computer Vision Engineer, AI Scientist, Research Scientist

Related on AIssential

• Deep learning-based detection of cessation of breathing in pre-term infants — Deep learning reliably detects infant apnoea using routine NICU signals, with signal modality being key.
• Two Stanford grads raise $11M to build a noninvasive wearable for hormone tracking — Clair Health offers a noninvasive wearable and AI for continuous, multi-biomarker hormone tracking to empower women's health understanding.
• Beyond Classification: A Cough Regression Benchmark for Respiratory Acoustic Foundation Models — Respiratory acoustic foundation models can predict continuous health quantities from cough audio using regression, moving beyond classification.
• Intelligent Healthcare Imaging Platform: A VLM-Based Framework for Automated Medical Image Analysis and Clinical Report Generation — A VLM-based framework automates medical image analysis and clinical report generation across multiple imaging modalities.
• Intervention-Based Self-Supervised Learning: A Causal Probe Paradigm for Remote Photoplethysmography — Intervention-based self-supervised learning can overcome correlation traps in rPPG by actively verifying physical signal hypotheses.
• State-Specific Respiratory Signatures for Affective and Stress Recognition: Interpretable Respiratory Markers, Autocorrelation Lags, and Compact CNN Models — Respiratory activity offers distinct, interpretable signatures for stress and affective state recognition, combining CNNs and handcrafted features.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Takara TLDR - Daily AI Papers.