Correlation analysis of respiratory signals by using parallel coordinate plots

Area of Science:

• Physiology
• Data Visualization
• Biomedical Engineering

Background:

• Understanding respiratory signals like airflow, mouth pressure, temperature, and humidity is crucial for improving respiratory mechanics measurements and diagnosing disorders.
• Existing methods for analyzing the complex interplay of these signals can be enhanced through novel data analysis approaches.

Purpose of the Study:

• To determine the relationships between multiple respiratory signals using a multidimensional data approach.
• To propose and evaluate a new combination of methods for analyzing respiratory signal coupling.

Main Methods:

• Employed statistical correlation analysis, including Pearson's correlation and cross-correlation coefficients.
• Utilized parallel coordinate plots (PCPs) with curve bundling, Least Mean Square Time Delay Estimator (LMS-TDE), and visual metrics for data interpretation.

Main Results:

• Correlation analysis revealed changes in signal coupling based on correlation coefficients, especially when analyzing whole breaths.
• LMS-TDE in PCPs corroborated correlation analysis findings for time delays.
• Visual metrics on PCPs identified signal ranges, dispersions, entropy, and linear/sinusoidal relationships.

Conclusions:

• The combination of correlation analysis and PCPs offers meaningful data views for user analysis.
• This approach can serve as an aid for researchers and clinicians in understanding respiratory signal dynamics.