INTERVAL ESTIMATION OF RESPIRATORY PARAMETERS USING LEAST-SQUARES TECHNIQUES

Relationships are determined that describe the bias inherent in point estimates of mean resistance and compliance obtained by applying least-squares to the respiratory equation P = RV + V C + P0 over inspiration and expiration, assuming flow follows a sinusoidal pattern. If resistance and elastance vary linearly over time, the bias can be described by linear functions of their respective ranges. The application of least-squares, using flow as the dependent variable, produces point estimates whose bias is no worse than when pressure is used. A least-squares technique is developed to obtain a piecewise-linear interval estimate of resistance and a linear-interval estimate of elastance over time, from pressure, flow, and volume data. This technique was applied to experimental data obtained on three subjects, with results indicating their resistance variation closely approximates a piecewise-linear pattern during both inspiration and expiration, while their elastance pattern appears linear in each phase. © 1979.