Passive sensing of gait and medication-related fluctuations in Parkinson's disease.

Only 15 people with Parkinson's and 14 controls took part, so the study was underpowered for anything beyond the largest effects. All Parkinson's participants had mild disease (UPDRS Part III of 32 or less), so the findings say nothing about moderate or advanced Parkinson's. The cohort was predominantly male (10 of 15 in the Parkinson's group, 10 of 14 in controls), limiting generalisability to women. The walk was a 4-metre instructed straight path in a home-like laboratory, not free-living gait in a real home, and the authors themselves flag this as the main limitation. The short walking distance means turning, acceleration, and continuous gait variability could not be assessed. Camera-derived asymmetry had a relative error of 72-88%, and the single-radar configuration could not distinguish left from right steps, so asymmetry detection from the radar was not possible.

Do not read the significant stride length differences as proof that these sensors can track all aspects of gait in Parkinson's. Only stride length survived the multiple-comparison correction; stride time, stride speed, and torso speed did not. The walk was a 4-metre instructed path in a lab-like room, not free-living gait in a real home.