Automated and accurate assessment for postural abnormalities in patients with Parkinson's disease based on Kinect and machine learning.
Zhuoyu Zhang, Ronghua Hong, Ao Lin and 9 others
PMID 34863184WHAT IT FOUND
A Kinect camera and machine learning system matched expert doctors' ratings of Parkinson's postural abnormalities with 90% accuracy.
It offers a low-cost, non-contact way to objectively quantify bent posture severity, potentially replacing subjective clinical scales.
Key findings
01The automated system achieved 90% accuracy in classifying postural abnormality severity, outperforming other tested algorithms.
02The agreement between the machine's score and the doctors' score was excellent (ICC 0.940), surpassing the agreement between the two human experts (ICC 0.814).
03A specific feature measuring the forward bulge of the back (F7) was the strongest predictor of severity, accounting for 40% of the model's decision-making.
STILL TO COME
How it was doneWhat they foundWhat it means for PTsWhat it means for OTs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study only assessed posture while standing still; it did not evaluate dynamic changes during walking. The system did not account for body torsion (twisting), which may lead to underestimating severity in some patients. The number of patients in each severity group was uneven, with very few in the most severe categories (6 patients in group '3' and 6 in group '4'). The study was conducted in a controlled setting with patients wearing close-fitting clothing, which may not reflect real-world clinic conditions where patients wear loose clothes. The system was validated against two doctors' ratings, but the 'gold standard' for postural assessment itself has known variability.
Declared interests
Funding was provided by several Chinese government science and technology programs. One funder is the 'clinical technology innovation project of allergan company', which sells botulinum toxin, a common treatment for dystonia and camptocormia. The study was also conducted in cooperation with iFLYTEK Suzhou Research Institute.
The easy way to misread this
Do not assume this system is ready for clinical use or that it works during movement. It was tested only on stationary standing postures in a controlled lab setting, and the most severe cases were represented by only 6 patients each, so its accuracy for extreme deformities or dynamic walking is unknown.