Related Experiment Video
Updated: Dec 16, 2025
08:22
Measuring Cardiac Autonomic Nervous System ANS Activity in Toddlers - Resting and Developmental Challenges
Published on: February 25, 2016
15.8K
Machine learning for automatic identification of thoracoabdominal asynchrony in children
Madhavi V Ratnagiri1, Lauren Ryan1, Abigail Strang2
1Biomedical Research, Nemours/Alfred I. duPont Hospital for Children, Wilmington, DE, USA.
Pediatric Research
|July 4, 2020
Summary
This study introduces an automated machine learning approach using the ICP feature to accurately identify thoracoabdominal asynchrony (TAA). This method significantly reduces diagnostic time and effort, improving upon traditional physician-dependent analyses.
Area of Science:
- Pulmonary Medicine
- Biomedical Engineering
- Machine Learning in Healthcare
Background:
- Current methods for assessing thoracoabdominal asynchrony (TAA) are time-consuming, requiring offline physician analysis or expert interpretation.
- Existing techniques like respiratory inductance plethysmography (RIP) and sleep apnea detection present limitations in efficiency and accessibility.
Purpose of the Study:
- To develop and validate an automated, accurate, and efficient method for identifying TAA.
- To explore the efficacy of machine learning (ML) with novel features for TAA assessment.
- To improve diagnostic consensus and reduce the burden on medical experts.
Main Methods:
- Utilized pneuRIP to measure thoracic and abdominal movements during quiet breathing.
- Trained an ML model with elastic-net regularization using phase difference (ɸ) and inverse cumulative percentage (ICP) features.
- Validated the model's performance on a separate dataset and compared outcomes with expert assessments.
Main Results:
- The ML model achieved 90.3% accuracy using the ICP feature, significantly higher than the 61.3% accuracy obtained with the ɸ feature.
- Inter-rater reliability among experts improved from 0.402 (using ɸ) to 0.684 (using ICP).
Conclusions:
- The ICP feature combined with ML provides an effective automated approach for TAA identification, reducing diagnostic time and effort.
- This automated method enhances diagnostic accuracy and improves expert consensus, offering a noninvasive and potentially remote monitoring solution.
