Classification of Parkinson's disease with freezing of gait based on 360° turning analysis using 36 kinematic features.

The study used laboratory 3D motion capture, so it does not show that a clinic can measure or use these features and cut-offs. The models were checked only by repeated internal splitting of the same data, not on a separate external group. The freezing classification sample was small and imbalanced, and the paper gives different freezer and non-freezer counts: 31 and 46 in the participant description, but 34 and 43 in the limitations. One freezer who froze during turning was excluded, so the analysis did not include an actual freezing episode. All testing was done in the Off medication state and only with the more affected limb on the inside of the turn, so On-state performance and other turning directions are unknown. Freezers had longer disease duration, higher levodopa equivalent dose, higher UPDRS total scores, and higher PIGD scores than non-freezers, so classification may reflect overall severity rather than freezing specifically. UPDRS part III did not differ between freezers and non-freezers, which makes the relationship between motor severity and freezing less straightforward. The regression models explained only a small amount of variation in turning characteristics, so associations should not be treated as strong clinical predictors. The study did not test any treatment or follow patients over time, so it cannot show that these turning features change falls, function, or quality of life.

The supplied text does not state authors' competing interests or funding. The journal metadata lists research support from non-U.S. government sources.

Do not treat the 98.1% Parkinson's disease result as a ready-made clinical test. It came from lab motion capture with internal cross-validation only, and the freezing versus non-freezing models reached 79.4% and 72.9%, with no external validation.