Table 1.
Representative studies examining remote monitoring of Parkinson’s disease (PD) motor symptoms using body worn sensors (BWS), between 2017 and 2020. Validation of digital outcomes has been classified ‘yes’ for each of the following criteria: 1. criterion validity: if digital outcome has been validated against a gold standard reference in the study cited, or in previous studies; 2. construct validity: if digital outcome has been validated (e.g., correlated) against clinical scales (convergent validity) and/or it has shown significant differences between groups (discriminant validity) in the study cited, or in previous studies. Digital outcome regulated/qualified has been classified ‘yes’ if BWS and/or digital outcome has received FDA (510K1) or EMA2 positive decision/qualification
| Study, Year | Dataset | Protocol | BWS Type/Position | Clinical Concept of Interest | Digital Outcome | Digital Outcome Validated (1. criterion validity, 2. construct validity) | BWS and/ or Digital Outcome Regulated/ Qualified |
| Tremor, bradykinesia, dyskinesia, motor fluctuations | |||||||
| Samà, et al., 2017 [13] | 12 PD | 1 day* (40 min), scripted tests ON and OFF state | IMUs (9 x 2)/Waist | Bradykinesia | Gait, frequency domain features for SVM model | 1. Yes, at home (SVM, against videos) | No |
| 2. Yes (convergent validity (UPDRS, UPDRS-III)) | |||||||
| Tsiouris et al., 2017 [14] | 20 PD | Scripted tests | PD manager: IMUs (Microsoft Band), Sensor insoles (Moticon), Smart Pillbox (SimpleMed+, Vaica), Smartphone/Wrist, Feet | Tremor, dyskinesia, bradykinesia, gait, FoG | Amplitude and constancy of tremor/not detailed features for ML techniques | 1. No 2. Yes (convergent validity (UPDRS)) | No |
| Wagner et al., 2017 [17] | 19 PD | 2 days | Accelerometer (GENEActiv)/Wrist | Tremor, bradykinesia, dyskinesia | Wavelet features (contribution and relative energy of each scale) | 1. Yes, in the lab (SVM, against clinician scores) | No |
| 2. No | |||||||
| Farzanehfar et al., 2018 [12] | 103 PD | 6-7 days, unsupervised | Accelerometer (PKG, Global Kinetic Corporation)/Wrist | Bradykinesia, dyskinesia | Bradykinesia score classified as movements with lower acceleration and amplitude. Dyskinesia classified as movements of normal amplitude and acceleration, but shorter periods without movement | 1. Yes, previous work [49] 2. Yes, (convergent validity (UPDRS III)) | Yes |
| Rodríguez-Molinero et al., 2018 [15] | 23 PD | 1–3 days | IMUs (9×2)/Waist | Bradykinetic gait, dyskinesia, ON-OFF state | Bradykinesia fluidity measure (frequency domain measure, power spectra 1–10Hz for each stride), dyskinesia (power spectra 1–4Hz for each stride) | 1. Yes (SVM, against diaries) 2. No | No |
| Rodríguez- Molinero, et al., 2019 [16] | 13 PD | 30 min of scripted activities | IMUs (9×2)/Waist | Dyskinesia | Power spectrum density in the frequency band comprised of harmonics of 1–4Hz | 1. Yes, in home environment using video data | No |
| 2. Yes, concurrent validity against clinical scales | |||||||
| Coates, et al., 2020 [46] | 5 PD, 5 OA | 7 days, unsupervised | Axivity AX3/Lower back | Motor symptom severity (MDS-UPDRS III) | Sample entropy (SampEnt) | 1. No | No |
| 2. Yes (convergent validity (against UPDRS III &levodopa equivalent daily dose (LEDD)) and discriminant validity (PD vs. CL)) | |||||||
| Evers et al., 2020 [11] | 25 PD, 25 CL | 1 day, scripted tests at home | IMUs (Gait Up Physilog 4, Android Wear smartwatch), contextual (smartphone) and physiological (Empatica E4) sensors/Lower back, wrists, ankles, pocket | ON-OFF state, FoG | Gait: Frequency domain measures (power spectral density (PSD), total power in the 0.5-10 Hz band, frequency, height and width of PSD dominant frequency) | 1. No 2. Yes (convergent validity (ON vs. OFF state) and discriminant validity (PD vs. CL)) |
No |
| Postural instability, gait disturbances, and turning | |||||||
| Rodríguez-Molinero et al., 2017 [26] | 75 PD | 1 day*, clinical assessment and scripted tests at home in ON and OFF state | IMUs (9×3)/Waist | UPDRS-III (axial function, balance, and gait) | Scalar value for ON-OFF state based on frequency domain features (power spectra 1–10Hz for each stride). | 1. Yes, in the lab, previous work, SVM against videos [50] 2. Yes (convergent validity (UPDRS-III, UPDRS-III factor 1: “axial function, balance, and gait.”)) | No |
| Haertner et al., 2018 [27] | 55 PD | 12 days (median) | IMUs (RehaGait®, Hasomed)/Lower back | Turning, falls risk | Duration, angle, average angular velocity, starting, middle and ending angular velocity and maximum angular velocity | 1. Yes, in home-like environment, previous work [51] | No |
| 2. Yes (discriminant validity (various PD fallers types)) | |||||||
| Mancini et al., 2018 [25] | 94 PD (25 freezers) | 3 days, unsupervised, clinical assessment and scripted test at home | IMUs (Dynaport Hybrid, McRoberts)/Lower back | Turning, FoG | Mean and coefficient of variation (CV) of: number of turns per 30 min, turn angle amplitude, turn duration, mean and peak turn velocity, turn jerkiness, turn medio-lateral range of acceleration. | 1. Yes, in the lab for turning, previous work [52] 2. Yes (convergent validity (NFOG-Q) and discriminant validity (freezers vs. non-freezers)). | No |
| Shah et al., 2020 [22] | 29 PD, 20 OA | 7 days, unsupervised* | IMUs (Opal, APDM)/Lower back, Feet | Gait | Gait speed, stride length, cadence, double-support, swing duration, pitch of feet at initial ground contact, frequency of bout length (number of strides) over a week | 1. Yes, in the lab, previous work [53] | No |
| 2. Yes (convergent validity (UPDRS-III, PIGD, previous work) and discriminant validity (PD vs. OA)) | |||||||
| Shah et al., 2020 [24] | 29 PD, 27 CL | 7 days, unsupervised* | IMUs (Opal, APDM)/Lower back, Feet | Gait, turning | 43 digital mobility characteristics (lower body, upper body, turning, activity, variability) | 1. Yes, in the lab, previous work [52, 53] 2. Yes (discriminant validity (PD vs. CL)) | No |
| Shah et al., 2020 [23] | 29 PD, 20 CL, 13 MS, 21 CL | 7 days, unsupervised* | IMUs (Opal, APDM)/Lower back, Feet | Gait, turning | 46 digital mobility characteristics (lower body, upper body, turning, activity, variability) | 1. Yes, in the lab, previous work [52, 53] | No |
| 2. Yes (convergent validity (UPDRS-III, PIGD) and discriminant validity (PD vs. CL)) | |||||||
| Falls risk, freezing of gait (FoG) | |||||||
| Rodríguez-Martín et al., 2017 [36] | 21 PD | 1 day* (40 mins), scripted tests ON and OFF state | IMUs (9×2)/Waist | FoG | 55 features for real-time SVM model | 1. Yes, at home (against videos) 2. No | No |
| Rodríguez-Martín et al., 2017 [37] | 12 PD | 3 days*, unsupervised, and in-lab scripted tests | IMUs (9×3)/Waist | FoG, bradykinetic gait | 55 features for real-time SVM model, frequency domain measures (strides) | 1. Yes, at home (against videos) 2. No | No |
| Mancini et al., 2018 [54] | 24 PD | 7 days*, unsupervised, clinical assessment and scripted test at home | IMUs (Opal, APDM)/Lower back, Ankles | FoG | Average of time spent freezing per hour (Total % time with Freezing ratio > 1 normalised on recording time), variability of % time spent freezing, turning and walking features | 1. Yes, in the lab, for turning, previous work [52] No for FoG and walking | No |
| 2. Yes for FoG, turning, and walking (convergent validity (NFOG-Q and ABC) and known group differences (freezers vs. non-freezers)). | |||||||
| Del Din et al., 2019 [29] | 155 PD F, 122 OA F, 15 PD NF, 50 OA NF | 7 days, unsupervised | Accelerometer, (AX3, Axivity)/Lower back | Falls risk, ambulatory activity | 14 Micro gait characteristics (pace, rhythm, variability, asymmetry, postural control), 7 Macro gait characteristics (volume, pattern, variability) | 1. Yes, in the laboratory for Micro gait characteristics, previous work [55]. In real-world for Macro gait characteristics in YA, previous work [56] | No |
| 2. Yes (convergent validity (FES-I), and discriminant validity (PD vs. OA, F vs. NF)) | |||||||
| Del Din et al., 2020 [30] | 128 PD F,109 OA F,38 MCI F | 7 days, unsupervised | Accelerometer, (AX3, Axivity)/Lower back | Falls risk, ambulatory activity | 7 Macro gait characteristics (volume, pattern, variability), fall rates relative to activity exposure (FRA) index | 1. Yes, in real-world for Macro gait characteristics in YA, previous work [56] | No |
| 2. Yes (convergent validity (FES-I), previous work, and discriminant validity (PD vs. OA)) | |||||||
| Reches et al., 2020 [38] | 71 PD | FoG provoking test in the lab in ON and OFF states | IMUs (Opal, APDM)/Lower back, Feet | FoG | 86 features from previous work for SVM model. | 1. Yes, in the lab against labelled video | No |
| 2. Yes (convergent validity (NFOG, UPDRS-III and TUG time), discriminant validity (OFF vs. ON state)) | |||||||
| Sigcha et al., 2020 [57] | 21 PD | 20 minutes at home, scripted ADLs in ON and OFF states | IMUs (9×2)/Waist | FoG | Mean, standard deviation, variance, frequency, entropy, energy, freeze index, sum of freeze index, locomotion band and variables related to FFT. ML and DL models from previous work. | 1. Yes, at-home against labelled video 2. No | No |
| Physical activity | |||||||
| Cai et al., 2017 [42] | 21 PD, 20 CL | 5 days, unsupervised | Bong Smart Sports bracelet/Wrist | Physical activity | Average daily physical activity amount and calories | 1. Yes, using self-report diaries 2. Yes (convergent validity (UPDRS –III, H&Y, Levodopa) and discriminant validity (PD vs. CL)) | No |
| Silva de Lima et al., 2018 [41] | 304 PD | 13 weeks | Pebble watch/Wrist | Physical activity/motor fluctuations (ON/OFF state) | Mean time spent walking | 1. No | No |
| 2. Yes (convergent validity (UPDRS item 4.4)) | |||||||
| Galperin, et al., 2019 [43] | 125 PD | 7 days, unsupervised | Accelerometer, Axivity (AX3)/Lower back | Physical activity | Acceleration derived features: Number of steps, number of walking bouts, step length, step regularity, amplitude of dominant frequency, SD of the peaks amplitude CV) and signal vector magnitude (SVM) | 1. Yes, in the laboratory against another BWS (GENEActiv which has been validated in previous work [58] 2. Yes (convergent validity (UPDRS-III)) | No |
| Pradhan et al., 2019 [40] | 30 PD, 30 OA | 14 days, unsupervised | Fitbit Charge HR/Wrist | Physical activity | Daily step count and METs | 1. Yes, in the laboratory and outdoor, previous work [59] | Yes (only ECG App) |
| 2. No | |||||||
| Ito et al., 2020 [44] | 13 PD | 1–7 days* | Accelerometer (Active Style Pro HJA 750C, OMRON)/Waist | Physical activity and motor symptoms (ON state, dyskinesia) | MET, PAL | 1. Yes in the laboratory, previous work [60] | No |
| 2. Yes (convergent validity (UPDRS-III, ON state, dyskinesia)) | |||||||
*Night excluded. ABC, Activity specific Balance Confidence scale; ADLs, Activities of Daily Living; CL, Controls; DL, Deep Learning; ECG, Electrocardiogram; F, Fallers; FES, Falls Efficacy Scale; H&Y, Hoehn and Yahr; MET, Metabolic Equivalent; ML, Machine Learning; MS, People with Multiple Sclerosis; NF, Non-fallers; NFOG-Q, New freezing of gait questionnaire; MDS-UPDRS, Movement Disorder Society Unified Parkinson’s disease Rating Scale; OA, Older Adults; PAL, Physical Activity Level; PD, People with Parkinson’s disease; PKG, Parkinson’s KinetiGraph; SVM, Super Vector Machine; TUG, Timed Up and Go; UPDRS-III, Unified Parkinson’s disease Rating Scale, Part III. 1https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm. 2https://www.ema.europa.eu/en/human-regulatory/research-development/scientific-advice-protocol-assistance/novel-methodologies-biomarkers/opinions-letters-support-qualification-novel-methodologies-medicine-development