Using smoothing splines for detecting ventilatory thresholds

Duane L. Sherrill, Stewart J. Anderson, George Swanson

Research output: Contribution to journalArticlepeer-review

17 Scopus citations


A recently developed nonparametric regression technique, called a polynomial smoothing spline, is presented for detecting the ventilatory threshold (VT) and respiratory compensation (RC) points from gas exchange response data taken during an incremental exercise test. This type of curve fitting has the advantage of not requiring investigators to specify, a priori, the form of the underlying model, as is required with all linear regression techniques. This procedure yields a mathematically optimal fitted curve through the O2 uptake (VO2) vs CO2 output (VCO2) data and estimates of the process’ first and second derivatives. A breakpoint or threshold is indicated by an increase in the value of the derivative and a peak in the second derivative. To evaluate this approach we analyzed gas exchange data collected on nine healthy subjects during a ramp exercise test (15 W·min−1) to the limits of tolerance. Utilizing this procedure we detected VT and RC breakpoints in four subjects and only VT breakpoints in the remaining five subjects. Our results and those obtained using the more conventinoal linear regression method were similar for those subjects whose RC points were detected using the smoothing spline procedure. However, using regression techniques for subjects with low or otherwise undetectable RC breakpoints, the linear regression method yielded less reliable results in our hands.

Original languageEnglish (US)
Pages (from-to)684-689
Number of pages6
JournalMedicine and Science in Sports and Exercise
Issue number5
StatePublished - Oct 1990
Externally publishedYes


  • Gas exchange
  • Incremental exercise
  • Lactate threshold

ASJC Scopus subject areas

  • Physical Therapy, Sports Therapy and Rehabilitation
  • Orthopedics and Sports Medicine


Dive into the research topics of 'Using smoothing splines for detecting ventilatory thresholds'. Together they form a unique fingerprint.

Cite this