Validation of a Revised V-Slope Algorithm for Automated First Ventilatory Threshold Detection in Healthy Adults.

Abstract

Background and objectives: The first ventilatory threshold (VT1) is a key indicator of submaximal aerobic fitness. Its determination typically relies on visual assessment, which is time-consuming and subject to interobserver variability. Automated solutions may offer a more objective, reproducible and time-efficient alternative. This study evaluated the accuracy of a revised V-slope method.

Methods: This retrospective study was conducted at University Hospitals Leuven, Belgium, using cardiopulmonary exercise test (CPET) data of 271 healthy adults collected between 2010 and 2020. Automated VT1 detection using a revised V-slope algorithm was compared with manual determination via the visual V-slope method in 262 participants with a determinable VT1. Agreement was evaluated using Bland-Altman analysis, intraclass correlation coefficients (ICC), Deming regression and equivalence testing (TOST). Interobserver agreement was evaluated using ICC.

Results: Median oxygen uptake at VT1 was higher with automated compared to manual assessment (1151 vs. 1122 mL/min; p < 0.001). Mean difference was 29 mL/min (95% CI [19, 40]. Relative mean bias was 2.2% ± 7.1% (LoA [- 11.7 to 16.0%], with excellent agreement (ICC 0.97 [95% CI 0.96-0.98] and a Deming regression slope of 0.95 [95% CI 0.92-0.99] and intercept of 26.1 mL/min [95% CI -11.9 to 64.7]. TOST with equivalence bounds [- 100, 100] was significant (p < 0.001; 90% CI [20, 38]).

Conclusion: The revised V-slope algorithm demonstrated strong group-level agreement with visual assessment for VT1 determination, supporting the reliability and practical applicability for large-scale research applications in healthy adults. However, given the observed individual-level variability, caution is warranted when applying automated VT1 values to guide exercise prescription for individual patients without expert review.