Athletes complete a balance of training loads and rest periods, risking overtraining when this balance favors excessive training. Diagnostic biomarkers have been suggested but a clear diagnostic method is not available. This preliminary study's objective was to use data standardization to improve an electronic nose's (enose) discrimination model for athletes' breathprints after cumulative and acute training loads.

Collegiate long distance runners were observed throughout competitive training seasons. Prolonged training effects were observed through Profile of Mood States (POMS) surveys and blood and breath samples collected at the beginning (Pre-Study) and end of the training season (Post-Study). Immediate training effects were observed for one low (LI) and one high (HI) intensity acute training load. Subjects provided blood and breath samples before the LI (BSR) and HI (BLR), completed the training load, and provided blood and breath samples after each training load (ASR; ALR). Blood was analyzed for antioxidant enzymes (catalase, glutathione peroxidase, and glutathione reductase). Breath samples were analyzed with a Cyranose® 320 (C320) enose.

Age, gender, and training loads affected oxidative states, with the HI having more effect than the LI. Mood profiles indicated healthy and successful athletes. Neither POMS nor blood parameters suggested overtrained athletes.

The C320 successfully discriminated between breathprints of athletes correlating to the training loads. Direct data standardization through carbon dioxide as a baseline sensor purge correctly classified 100 percent of the data through linear discriminant analysis (LDA). Indirect data standardization by subtracting Pre-Study data from the subsequent data classes (e.g. BSR) correctly classified 96 percent of the data.

An LDA on the combined blood parameters correctly classified 61.9 percent of the data. The blood analyses required invasive sample collections and involved procedures that took a long time (hours). In comparison, the best C320 model correctly classified 96 percent of the data and required less invasive sample collections, simple analysis, and short result times (minutes).

Evidence suggested the C320 will provide a simple and noninvasive method for clinically diagnosing the onset of overtraining. The unit is small, handheld, rapid, and noninvasive so it could also be used on- site to provide immediate feedback for training optimization.