An Automated Application For Analysis Of Incremental Exercise Tests

Introduction Clinical exercise tests (CXT) using symptom-limited incremental protocols can identify physiological limitations and assist with diagnosis of exercise intolerance. However, there are challenges with these tests because of lack of standardization in their performance and with data interpretation. Methods We developed an automated application for CXT interpretation based on novel algorithms and pattern recognition. We tested the application on 100 sequential CXT (57 men) and compared results with human interpretation by experienced physicians. Six key physiological variables were examined: maximum values for work (Wmax), oxygen uptake (VO 2 max), heart rate (fCmax), and ventilation (VEmax); and slope values for chronotropic index (CI) and ventilatory efficiency (VE). Accuracy was compared to stringent minimum clinically important differences. Agreement was assessed by Bland-Altman methods. Results Three studies were excluded from the analysis because of incomplete data. In three more, the heart rate signal was corrupted by interference. Accuracy and mean bias (limits of agreement) were: Wmax 96%, 1.7 (5.6, -2.2) watts, VO 2 max 80%, -0.02 (0.13, -0.18) L/min, fCmax 82%, -2.0 (8.1, -12.2) /min, VEmax 75%, 3.1 (13.7, -19.85), CI 79%, -0.7 (11.6, -10.3), VE 69%, 1.7 (16.5, -13.1). The application also provides plots for visual detection of threshold values for lactate accumulation (VO 2 θ) and ventilatory compensation (VCO 2 θ). Conclusion This automated application produces more consistent and reliable characterization of the human exercise response than manual reporting. Such standardization is necessary to better define the pathophysiology and to assist with diagnosis of exercise limitation.