Multivariable Regression Model to Estimate Tidal Volume for Different Respiratory Patterns.

Area of Science:

• Respiratory Physiology
• Biomedical Engineering
• Data Analysis

Background:

• Respiratory patterns exhibit significant variability in healthy individuals and patients.
• Accurate analysis of respiratory variability relies on appropriate recording devices and signal processing.
• Existing methods for tidal volume estimation may not fully capture the complexity of diverse breathing patterns.

Purpose of the Study:

• To propose and validate multivariable regression models for estimating tidal volume (V_T) considering different respiratory patterns.
• To compare the performance of models using thoracic versus abdominal respiratory inductive plethysmography bands.
• To assess the improvement in V_T estimation by combining parameters from multiple sensors and analyzing specific breathing patterns.

Main Methods:

• Twenty-three healthy volunteers underwent continuous multisensor recordings during various breathing modes.
• Respiratory flow and volume signals were captured using a pneumotachograph and respiratory inductive plethysmography bands (thoracic and abdominal).
• Multivariable regression models were developed using extracted respiratory parameters (e.g., areas, times, volumes) to estimate V_T, with performance evaluated by R², E_R, and IQR.

Main Results:

• Models using three thoracic band parameters (VT_exp, T_tot, Area_exp) yielded better V_T estimation (R² = 0.94) than abdominal band parameters (R² = 0.91).
• Combining parameters from both thoracic and abdominal bands significantly improved overall V_T estimation (R² = 0.97).
• Specific breathing patterns, such as nose-nose, nose-mouth, and mouth-mouth, showed superior estimation accuracy compared to basal, shallow, or deep breathing.

Conclusions:

• Multivariable regression models effectively describe respiratory pattern variability and estimate tidal volume.
• Integrating data from multiple respiratory inductive plethysmography bands enhances the accuracy of V_T estimation.
• The developed models demonstrate clinical relevance for analyzing diverse breathing patterns and improving respiratory monitoring.