Automated quantitative analysis of capnogram shape for COPD-normal and COPD-CHF classification.

Area of Science:

• Pulmonary Medicine
• Biomedical Engineering
• Data Science

Background:

• Capnography provides respiratory data, but its waveform shape is qualitatively assessed.
• Quantitative capnogram analysis for disease diagnosis has seen limited clinical translation.
• Existing methods struggle to consistently characterize capnogram shape for diagnostic value.

Purpose of the Study:

• To develop an automated quantitative approach for analyzing capnogram shape.
• To discriminate between chronic obstructive pulmonary disease (COPD) and congestive heart failure (CHF).
• To differentiate between COPD and normal subjects using capnogram features.

Main Methods:

• Collected capnograms from 30 normal subjects, 56 COPD patients, and 53 CHF patients.
• Computationally extracted four physiologically-based capnogram features.
• Employed an ensemble of classifiers with quadratic discriminant analysis for classification.

Main Results:

• Achieved an area under the ROC curve of 0.89 for COPD/CHF classification.
• Achieved an area under the ROC curve of 0.98 for COPD/normal classification.
• Demonstrated a fast run time suitable for real-time monitoring.

Conclusions:

• Automated quantitative capnogram analysis can effectively distinguish between COPD, CHF, and normal individuals.
• This approach offers a promising tool for objective diagnosis and monitoring of respiratory conditions.
• The method's speed and accuracy support potential clinical integration for real-time patient assessment.