Freezing of gait assessment with inertial measurement units and deep learning: effect of tasks, medication states, and stops

Table 1 Naming convention of the deep learning models evaluated in this study with their corresponding training data

Usage	Model name	FOG-provoking task		Medication state		Stopping
Usage	Model name	TUG	360Turn	Off	On	Exclude	Include
FOG detection in clinical practice	Model_TUG	v		v	v	v
	Model_360Turn		v	v	v	v
	Model_Off	v	v	v		v
	Model_On	v	v		v	v
	Model_Clinical	v	v	v	v	v
Towards FOG detection in daily life	Model_Stop	v	v	v	v		v

Models trained for specific tasks or medication states for standardized measurements were termed “Model_TUG”, “Model_360Turn”, “Model_Off”, and “Model_On”. A DL model trained for FOG assessment in the clinical centers, trained on standardized tasks excluding stops, was termed “Model_Clinical”. A DL model trained to work towards FOG assessment in daily life, trained on standardized tasks including stops, was termed “Model_Stop”

ISSN: 1743-0003