Improved Calibration Method of Respiratory Belts by Extension of Multiple Linear Regression

Respiratory belts are widely used for ventilation monitoring. After calibration they can be used quantitatively to measure breathing volume and flow. This study provides an improved calibration method for respiratory belt signals. An optimized FIR (finite impulse response) filter bank was constructed as a MISO system (multiple input, single output) between the respiratory belts and spirometer. First-order and second-order polynomials with cross-terms were tested in the transfer function. Six subjects were measured with two respiratory belts (piezo and inductive) at the same time. Results show that R-2 value increased (piezo belts: 9%, inductive belts: 10%) and RMSE (root mean square error) decreased greatly (piezo belts: 36%, inductive belts: 43%) compared to the golden standard calibration method. It is concluded that the new method improves greatly the accuracy of calibration of respiratory belts.