Table 4.

Technical characteristics of wearable devices.

Study, year	Sensor	Wearable device type	Placement of wearable device	Measured biosignal	Connectivity	Host device	Sensing approach	AIa and MLb technology used	Clinical aim of AI and ML technology
Buckley et al [74], 2019	Accelerometer, gyroscope, and magnetometer	Inertial measurement unit	Head, lumbar region, and pelvis	Upper body acceleration and spatiotemporal characteristics	Wi-Fi	Data server	Opportunistic	Receiver operating characteristic and area under the curve	Determine the correlation of clinical feature and gait analysis
Coates et al [75], 2020	Accelerometer, light sensor, and temperature sensor	Smart band	Lumbar region	Sample entropy and gait	USB cable	Data server	Opportunistic	IAAFTc	Determine the correlation of clinical features
De Vos et al [76], 2020	Accelerometer, gyroscope, and magnetometer	Inertial measurement unit	Lumbar region, wrist, and feet	Gait and postural sway	Wi-Fi	Data server	Opportunistic	LASSOd and random forest	Parkinson disease classification
Dominey et al [77], 2020	Accelerometer	Smartwatch	Wrist	Bradykinesia, dyskinesia, and sleep disturbance	Wi-Fi	Data server	Opportunistic	No	—e
Dunne-Willows et al [78], 2019	Accelerometer, light sensor, and temperature sensor	Smart band	Lumbar region	Gait	USB cable	Data server	Opportunistic	Phase algorithm	Gait analysis
Huo et al [69], 2020	Force sensor, accelerometer, gyroscope, magnetometer, and mechanomyography sensor	Smart band	Arm	Rigidity, tremor, and bradykinesia	Bluetooth	Data server	Opportunistic	1-NNf, AdaBoost classifier, and MLPg neural networks classifier	Ranking Parkinson disease symptoms
Lacy et al [79], 2018	Electromagnetic sensor	Smart finger	Finger	Finger tapping	Sensor cable	Data server	Opportunistic	Echo state network	Parkinson disease classification
Lones et al [80], 2017	Accelerometer, gyroscope	Smart band	Head, chest, arms, and legs	Dyskinesia	Wi-Fi	Mobile phone	Opportunistic	IRCGPh	Parkinson disease classification
Mc Ardle et al [72], 2020	Accelerometer, light sensor, and temperature sensor	Smart band	Lumbar region	Gait	USB cable	Data server	Opportunistic	Black box	Parkinson disease classification
Mc Ardle et al [73], 2021	Accelerometer, light sensor, and temperature sensor	Smart band	Lumbar region	Gait and balance impairment	USB cable	Data server	Opportunistic	Area under the curve	Parkinson disease classification
Pantall et al [81], 2018	Accelerometer, light sensor, and temperature sensor	Smart band	Lumbar region	Sample entropy	USB cable	Data server	Opportunistic	Area under the curve	Parkinson disease classification
Pantall et al [82], 2018	Accelerometer, light sensor, and temperature sensor	Smart band	Unreported	Balance impairment and posture	USB cable	Data server	Opportunistic	Black box	Determine the correlation of clinical features
Payne et al [70], 2020	Accelerometer, gyroscope, and magnetometer	Inertial measurement unit and Smart band	Head, upper back, lumbar region, and ankles	Motor impairment and gait	Wi-Fi	Data server	Opportunistic	Area under the curve	Determine the correlation of clinical features
Rehman et al [83], 2019	Accelerometer, light sensor, and temperature sensor	Smart band	Lumbar region	Gait	USB cable	Data server	Opportunistic	Support vector machine and random forest	Gait analysis and Parkinson disease classification
Rehman et al [84], 2020	Accelerometer, light sensor, temperature sensor	Smart band	Lumbar region	Gait	USB cable	Data server	Opportunistic	PLS-DAi	Gait analysis and Parkinson disease classification
Rehman et al [85], 2020	Accelerometer, gyroscope, and magnetometer	Inertial measurement unit	Head, neck, lumbar region, and ankles	Gait	Wi-Fi	Data server	Opportunistic	PLS-DA	Gait analysis Parkinson disease classification
Stack et al [86], 2018	Accelerometer and gyroscope	Smart band	Wrist	Instability	Unreported	Unreported	Opportunistic	No	—
Stuart et al [87], 2015	Accelerometer, light sensor, temperature sensor, and eye tracker	Smart band and Smart glasses	Head	Eye movements and gait	USB cable and Wi-Fi	Data server	Opportunistic	No	—
van Wamelen et al [88], 2019	Accelerometer	Smartwatch	Wrist	Bradykinesia, dyskinesia, and levodopa equivalent dose	Wi-Fi	Data server	Opportunistic	Black box	Determine the correlation of clinical features
van Wamelen et al [89], 2021	Accelerometer	Smartwatch	Wrist	Rigidity, tremor, and bradykinesia	Wi-Fi	Data server	Opportunistic	Black box	Determine the correlation of clinical features

^aAI: artificial intelligence.

^bML: machine learning.

^cIAAFT: iterated amplitude–adjusted Fourier transform.

^dLASSO: least absolute shrinkage and selection operator.

^eNot available.

^fNN: nearest neighbor.

^gMLP: multilayer perceptron.

^hIRCGP: implicit context representation Cartesian genetic programming.

ⁱPLS-DA: partial least square discriminant analysis.