Regresar

Estimation of invasive physiological parameters from non invasive parameters using dimensionless numbers and Monte Carlo cross-validation

Abstract:

According to NCEP-ATPIII criterion, metabolic syndrome (MS) diagnosis is based in the measurement of invasive (triglycerides, HDL, glucose) and non invasive variables (height, weight, waist circumference). The aim of this work is find, since dimensionless numbers design from physiological (π1IS, π2IS) and heart rate variability (π1HRV, π2HRV) parameters, three polynomial equations (π1IS=f(π2HRV), π2IS=f(π1HRV), π2IS=f(π2HRV)) that relate invasive with non invasive variables. In this sense, a fitting using Monte Carlo cross validation (MCCV) was performed. A database of 40 subjects (25 control subjects and 15 subjects with MS) was employed. The results suggest that MCCV improves the coefficient of determination (R2), compared to the application of the least squares method only, in each polynomial: π1IS=f(π2HRV) (R2=0.62 vs. 0.21), π2IS=f(π1HRV) (R2=0.62 vs. 0.36) and π2IS=f(π2HRV) (R2=0.56 vs. 0.19). The fitting by MCCV allows the estimation of invasive from non invasive parameters.