Estimation of Respiratory Displacements Using a Nonlinear Observer

This paper develops and evaluates a wearable sensor-based system for the estimation of three-dimensional thoracoabdominal displacements. Such estimation is useful for a number of respiratory diagnosis applications, including detection of paradoxical breathing, quantification of tidal volume, and determining efficiency of mechanical ventilation. The use of inertial sensors on the body to estimate respiratory displacements is challenging due to the tilting of the body that occurs with breathing, causing continuous changes in the gravity component at the same frequency as the breathing frequency. A method to estimate the front-to-back and side-to-side tilt angles is developed using a nonlinear observer. The global stability of the observer is analyzed using Lyapunov analysis. Example experimental results are presented using data from a supine subject at various breathing frequencies and tidal volumes. Displacement estimates are found to be typically within ±1 mm compared to a gold standard reference for most data sets.