AI-Techniques Loss-Based Algorithm for Severity Classification (ATLAS): a novel approach for continuous quantification of exertional symptoms during incremental exercise testing

Abstract

Objective Heightened muscular effort and breathlessness (dyspnea) are disabling sensory experiences. We sought to improve the current approach of assessing these symptoms only at the maximal effort to new paradigms based on their continuous quantification throughout cardiopulmonary exercise testing (CPET). Materials and Methods After establishing sex- and age-adjusted reference centiles (0-10 Borg scale), we developed a novel algorithm ( A I T echniques L oss-Based A lgorithm for S everity Classification [ATLAS]) based on reciprocal exponential loss for CPET data from patients with chronic obstructive lung disease of varied severity. Results Categories of dyspnea intensity by ATLAS—but not dyspnea at peak exercise—correctly discriminated patients in progressively higher resting and exercise impairment (P