Moving beyond threshold-based dichotomous classification to improve the accuracy in classifying non-responders

Area of Science:

• Exercise Physiology
• Sports Science
• Cardiorespiratory Fitness

Background:

• Maximal oxygen consumption (VO2 max) is a key indicator of cardiorespiratory fitness.
• Exercise training aims to improve VO2 max, but individual responses vary significantly.
• Current classification methods often use arbitrary thresholds, leading to inaccurate responder/non-responder categorization.

Purpose of the Study:

• To evaluate the limitations of threshold-based classification for exercise training responders.
• To propose and validate alternative methods for classifying individual responses to exercise.
• To enhance the accuracy of identifying individuals who do not benefit from training interventions.

Main Methods:

• Calculated individual probabilities of response to exercise training.
• Classified individuals using response confidence intervals (CI).
• Utilized reference points including zero change and a smallest worthwhile change (0.5 METs).

Main Results:

• Threshold-based classification presents significant limitations in accurately identifying responders and non-responders.
• Individual probabilities of response provide a more nuanced understanding of training effects.
• Individual confidence intervals, when applied with appropriate reference points, improve classification accuracy, particularly for non-responders.

Conclusions:

• The dichotomous classification of exercise training responders and non-responders based on fixed thresholds is inadequate.
• Individualized assessment using response probabilities and confidence intervals is superior for accurate classification.
• These advanced methods improve the identification of non-response to exercise, aiding in personalized training prescription.