Automated and accurate assessment for postural abnormalities in patients with Parkinson’s disease based on Kinect and machine learning

Table 3 Selection results of the relevant features

	ICC_DD	ICC_MD
F1 + F2 + F3 + F4 + F5 + F6 (max depth = 5, splitter = ‘best’)	0.814	0.707
F1 + F2 + F3 + F4 + F5 + F7 (max depth = 5, splitter = ‘best’)	0.814	0.786
F1 + F2 + F3 + F4 + F5 + F6 + F7 + F8 (max depth = 5, splitter = ‘best’)	0.814	0.781
F1 + F2 + F3 + F4 + F5 + F7 (max depth = 7, splitter = ‘random’)	0.814	0.940

ISSN: 1743-0003