Principal component analysis detects sleepiness-related changes in balance control

Computerized posturography exploits balance scores that quantify the size, dynamics, or structure of the recorded sway. Since people employ different balance strategies, one single balance score will not detect balance changes in all subjects. Principal component analysis (PCA) can combine balance scores that quantify different features into one new balance score. We tested the score with 20 subjects by measuring their balance every 2 h during 28 h of sustained waking. The new balance score was more sensitive than its components (p<0.001 vs. p≥0.051) to the small sleepiness-related balance decrements that occurred during the short 28-h period. PCA provided a more sensitive balance score that applied to all of the subjects.