https://orcid.org/0000-0003-3363-3077
Figures
Abstract
Background
Wearable sensors-based systems have emerged as a potential tool to continuously monitor Parkinson’s Disease (PD) motor features in free-living environments.
Objectives
To analyse the responsivity of wearable inertial sensor (WIS) measures (On/Off-Time, dyskinesia, freezing of gait (FoG) and gait parameters) after treatment adjustments. We also aim to study the ability of the sensor in the detection of MF, dyskinesia, FoG and the percentage of Off-Time, under ambulatory conditions of use.
Methods
We conducted an observational, open-label study. PD patients wore a validated WIS (STAT-ONTM) for one week (before treatment), and one week, three months after therapeutic changes. The patients were analyzed into two groups according to whether treatment changes had been indicated or not.
Results
Thirty-nine PD patients were included in the study (PD duration 8 ± 3.5 years). Treatment changes were made in 29 patients (85%). When comparing the two groups (treatment intervention vs no intervention), the WIS detected significant changes in the mean percentage of Off-Time (p = 0.007), the mean percentage of On-Time (p = 0.002), the number of steps (p = 0.008) and the gait fluidity (p = 0.004). The mean percentage of Off-Time among the patients who decreased their Off-Time (79% of patients) was -7.54 ± 5.26. The mean percentage of On-Time among the patients that increased their On-Time (59% of patients) was 8.9 ± 6.46. The Spearman correlation between the mean fluidity of the stride and the UPDRS-III- Factor I was 0.6 (p = <0.001). The system detected motor fluctuations (MF) in thirty-seven patients (95%), whilst dyskinesia and FoG were detected in fifteen (41%), and nine PD patients (23%), respectively. However, the kappa agreement analysis between the UPDRS-IV/clinical interview and the sensor was 0.089 for MF, 0.318 for dyskinesia and 0.481 for FoG.
Citation: Caballol N, Bayés À, Prats A, Martín-Baranera M, Quispe P (2023) Feasibility of a wearable inertial sensor to assess motor complications and treatment in Parkinson’s disease. PLoS ONE 18(2): e0279910. https://doi.org/10.1371/journal.pone.0279910
Editor: Keisuke Suzuki, Dokkyo Medical University: Dokkyo Ika Daigaku, JAPAN
Received: May 2, 2022; Accepted: December 18, 2022; Published: February 2, 2023
Copyright: © 2023 Caballol et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are available upon request, due to ethical restrictions imposed by Spanish Agency for Drugs and Medical Devices and the Local Clinical Research Ethics Committee, related to approved consent procedure and protecting privacy. Data can be available upon request to qs.investigacion@quironsalud.es.
Funding: The author(s) received no specific funding for this work.
Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: AB is shareholder of Sense4Care. This does not alter our adherence to PLOS ONE policies on sharing data and materials. AB reports receiving honoraria from Bial and Zambon. NC reports receiving consultancy fees as member of the advisory board from Italfármaco, Abbvie and Zambon, speaking fees from Lundbeck and Zambon, and honoraria from Bial, Zambon and UCB. AP declares no conflict of interest. PQ declares no conflict of interest. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
Introduction
Parkinson’s disease (PD) is a complex neurodegenerative disorder characterized by a wide range of motor and non-motor symptoms. Dopaminergic treatments can improve symptoms and quality of life. However, motor fluctuations (MF), non-motor fluctuations (NMF), dyskinesia and freezing of gait (FoG) complicate the management of PD [1–3]. Accurate identification of motor symptoms, motor fluctuations and Off-Time is crucial to making a precise treatment adjustment. The clinical interview, a set of recommended clinical scales and patient’s diaries are widely used for this purpose [4–7]. Nonetheless, the real time for each patient in the setting of the clinical practice is sometimes limited and not all the symptoms can be captured properly. Besides, intra and interrater correlations of UPDRS are low and give only a “quick snap picture” of the motor state of the patient [8]. Completing patient’s diaries require some effort, but they are completed without difficulty in most instances [9]. However, some patients may struggle to register motor complications (MC) accurately. Recall bias, reduced diaries compliance and diary fatigue can also be common [10,11].
New technologies and wearable sensors-based systems have emerged in the last decade to objectively assess motor PD symptoms for long periods in free-living environments [12–19]. Novel devices based either on machine-learning approaches or statistical-based methodology, can be used to capture and monitor motor PD symptoms and MC, or to objectively assess response to dopaminergic therapy [20–33]. The Kinesia 360TM is composed of two sensors, wrist-worn and ankle-worn [32]. It uses a gyroscope, and the algorithms give information regarding tremor, dyskinesia, slowness, mobility, posture and steps. The PKGTM has been extensively used. It is a waist-worn sensor that can detects ON/OFF states, bradykinesia, dyskinesia, tremor, and inactivity state but not FoG, gait or falls [26–30]. The PDMonitorTM uses a five-device system and can detect ON/OFF MF, bradykinesia, dyskinesia, tremor, FoG, gait measures and rest state but not falls. Although these novel tools are promising, some challenges such as the lack of external validation could limit their implementation [8]. The main challenge relies on comparing the clinical practice with devices against the clinical practice without devices and seeing if the wearable sensor-based systems can improve PD motor symptoms. So far, only comparisons between sensors and questionnaires have been done, but no clear evidence has been shown yet.
The STAT-ONTM is a wearable inertial sensor (WIS) that was certified as a medical device in June 2019. The system provides numerical and graphical information of the motor symptoms associated with PD. The STAT-ON reports include the percentages of Off-Time, On-Time, dyskinesia, and FoG (Fig 1A and 1C). The WIS can capture On, Off or intermediate states (Fig 1B). When no movement is detected, the report shows this period in grey colour (Fig 1B). The sensor also captures the gait fluidity (GF) (Fig 1D). The minutes walked, the number of steps per day, the cadence and the number of falls can also be detected.
A. Example of the first page of the STAT-ONTM report. It contains information regarding the total time of monitorization along the 7-day-period. B. Example of the report that is shared with the patient. Color legend. Green: On state; Red: Off state; Yellow: Intermediate state; Grey: Inactivity or rest state. No walking detection; Freezing of Gait (FoG) is represented by a black round; Horizontal purple line: Dyskinesia; Vertical black line: Button event (in the present study, corresponds with the levodopa dose); Vertical blue line: Fall. C. The figure shows the total time and the percentage of Off state of every day. The height of each bar depends on the percentage of Off-Time per day. D. This figure shows the GF along the day of the third day of monitoring of the same patient of the STAT-ONTM report above. The GF is worse in the morning.
A recent study that analysed the opinion of different PD experts in Spain regarding WIS clinical utility showed that the system was considered “quite” to “very useful” by 74% of the neurologists [34]. Moreover, a single-blind, multicentre, randomised controlled clinical trial comparing the effectiveness of the system with other clinical monitoring methods is being conducted [35]. Satisfaction and usability of the system among PD patients have been also analysed using the Quebec User Evaluation of Satisfaction with assistive Technology (QUEST) questionnaire and the System Usability Scale (SUS) questionnaire respectively [36–38]. The level of satisfaction of the device was high according to QUEST scale (all items scored between 4 “quite satisfied” and 5 “very satisfied”), except for the item “easy in adjusting” (47%), which had a lower score [38].
Regarding the sensitivity and specificity of the algorithms embedded into the sensor for detecting On-Off fluctuations, dyskinesia or FoG, a set of studies have demonstrated that both are around 0.9, with few variations depending on the algorithm [39–47]. However, there has been no published data regarding the responsivity of the sensor’s measures after treatment adjustments. Besides, and to the best of our knowledge there are no studies analysing the agreement of the detection of MF, dyskinesia, and FoG in an unsupervised scenario.
We conducted an observational, open-label study in our outpatient Movement Disorders Unit to assess this both issues of the sensor, under ambulatory conditions of use.
Methods
Study participants and inclusion criteria
Patients diagnosed with PD according to the UK PD Society Brain Bank criteria who routinely visited our unit, were recruited prospectively between September 2019 and December 2019 [48]. All participants met the following inclusion criteria: aged 30–80 years, were taking levodopa, had MC or a “suboptimal” motor state, and agreed to participate. A suboptimal motor state was considered when MF or WO were not clearly manifested according to the patient and the UPDRS-III motor scale of the patient was 3 points worse than the evaluation in the previous visit (the change considered by the literature as clinically significant). Participants were excluded if it was deemed by the interviewer’s clinical judgment that they were unable to provide valid responses. Exclusion criteria also included unavailability to perform the study because of hospitalization or diagnosis of serious illness.
Design, procedures, and clinical assessment
Socio-demographic and clinical data were collected. Standardized neurological assessment included the clinical interview, the Unified Parkinson’s Disease Rating Scale (UPDRS)- II, UPDRS-III, UPDRS-IV, and the H&Y Staging of PD.
Patients were trained on how to use the device. The WIS was placed on their left hip in the first visit, and for one week, twelve hours a day (Fig 2).
After one week, the sensor was returned and WIS information was evaluated jointly by the neurologist and the patient. Treatment adjustments were indicated mainly according to best clinical practice guidelines but in some cases, device information reinforced these treatment decisions. The patients were analysed into two groups according to whether treatment changes had been indicated or not. In addition, the WIS satisfaction through QUEST scale was evaluated.
After 11 weeks, the WIS was again delivered for patients to use for a 7-day period, twelve hours per day. WIS information was shared with the patient. Assessment of FoG, Off-Time, and UPDRS II-IV and the SUS questionnaire were also completed (Fig 2).
Patient’s training on how to use the sensor
Patients were asked to place the WIS on their left hip during the day for 12 hours. They were encouraged to continue with their usual activities as much as possible. For a better understanding of the WIS information, patients were asked to record their ADL each hour. However, we asked them to take off de WIS if a long car journey was done, or they practiced physical activity. They were trained to push the unique button of the system when they took a levodopa dosage.
Equipment
We used the STAT-ONTM system (https://www.statonholter.com/) as a continuous monitoring system through new technologies. The system is a Medical Device Class IIa (Council Directive 93/42/EEC of 14 June 1993) being only prescribed and provided from the clinician to the patient. The STAT-ONTM system consists of a monitoring device, its base charger, a belt, and a smartphone application (App). According to the Instructions for Use, The STAT-ONTM is a waist-worn inertial recorder, configured by a doctor and used by the patient for clinical, ambulatory, or home environments, that collects the results of the motor disorders and events of PD patients for a period. The smartphone can be property of the neurologist or the centre. The device continuously collects the inertial signals of the patient’s movement, processes them in real-time by means of artificial intelligence algorithms, and stores the results in its internal memory. When the patient returns the sensor, the App connects to the STAT-ONTM device via Bluetooth and WIS data are downloaded and converted into structured report that can be emailed to any user or neurologist (Fig 1A–1D).
The App can only be managed by the health staff and can be installed on any smartphone or tablet (Android 5 or higher) or iOS for Apple devices (iOS 10.2 or higher). STAT-ONTM App is used both for configuring the system and for downloading the data previously generated by the sensor. The distance between the ground and the user’s hip (leg length), the age and the H&Y Off stage are needed for a use configuration.
Assessment of motor fluctuations, dyskinesia, FoG and patient-based Off-Time
Motor fluctuations were assessed by UPDRS-IV (item 39) while assessment of dyskinesia was based on item 35 of UPDRS-IV. The presence of FoG was evaluated according to the clinical interview.
Data WIS analysis
Selection of monitored days.
The analysis was performed selecting the 4 days with more hours in Off-Time. Although patients were asked to wear the sensor for 12 hours, the time of use during the day was variable. Therefore, all those days with less than 8 hours of monitoring were omitted. The variation of having days with 8 to more than 12 hours of monitoring was minimized using the percentage of time in OFF with respect to the total time monitored. Besides, in our experience there are some patients with fewer hours monitored that put on the sensor later in the morning. This can lead to missing data of the morning-Off’s. The most appropriate and simple criterion to minimize this effect and not introduce bias in the sample was to compare the 4 worst days of each patient in terms of the percentage of OFF with respect to the monitored time.
Analysis of sensor data for the correlation with clinical data.
The outcome rate of the sensor’s information is provided every 30 minutes. This means that if 12 hours of monitoring are captured per day for 7 days, a total of 168 outcomes are obtained. Obviously, this is not directly comparable with a single clinical evaluation. In order to obtain a single diagnostic data for all the monitoring, different criteria and aggregations have been applied, which we detail below. In our experience, the most natural way to aggregate the sensor information is through a timeframe of one day. This means that the percentage of the different variables (Off, On, Intermediate- Time or dyskinesia) have only been computed if the monitoring period is more than 8h per day. From these daily measurements, we analyse the 4 worst days with higher OFF percentages. Finally, and with the aim of normalising the data amongst the patients, from these obtained 4 days, we calculate the mean of all the parameters. This way, it is obtained in a single data diagnosis which could be comparable to the clinical evaluation carried out in the consultation.
Factor I of UPDRS-III.
UPDRS-III-Factor I is one of the parameters of the UPDRS that correlates better with the sensor signals. The UPDRS-III items clustered in the Factor-I are speech, facial expression, arising from a chair, gait, postural instability, and body bradykinesia [39,46,49]. The fluidity of the stride gait is a continuous variable, calculated for each group of strides of the patient, and which is the basis for calculating motor states [50–52]. In the present study we analysed the correlation of the Factor I and the mean fluidity of the stride. This analysis was evaluated in all the patients that completed the two monitorizations (after and before treatment changes).
Statistical methods
Sensor variables were measured at baseline and three months later. The changes in these variables were compared between the patients in whom some therapeutic intervention was performed versus those that did not. Univariate comparisons were performed between both groups. The statistical tests included Chi-squared test, U-Mann-Whitney test, and t-student test as appropriate. Spearman correlation was applied for the analysis of Factor I of UPDRS-III. For the agreement analysis of MF, dyskinesias and FoG, the proportion of observed agreement and the Kappa coefficient was estimated for every pair of observers and every symptom. Statistical significance was set at p<0.05. SPSS 24.0 statistical software (SPSS, Chicago, IL) was used for all statistical analyses.
Results
Characteristics of PD sample and WIS parameters at baseline
Thirty-nine PD patients were included in the study [mean (SD) age of 69 (8) years, 56% men, mean disease duration PD 8 (3.5) years]. All PD patients were treated with levodopa. Twenty-nine PD patients (74%) had predictable MF according to UPDRS-IV Q36, while 7.7% had unpredictable MF. Seventeen patients (44%) had dyskinesia and twelve patients (31%) had FoG according to the clinical interview. Details regarding, demographics, motor state and therapies of sample, according the two groups (therapeutic intervention vs not) are shown in Table 1.
The WIS captured On, Off and intermediate state in the majority of patients (95–97%).
Characteristics regarding the motor state, the MC and patients’ gait parameters according to the sensor are detailed in Table 2.
At the 3-month follow-up visit, five dropouts (13%) occurred: in three patients due to intercurrent diseases and in two patients due to unavailability to perform the study, all of them not related to PD.
Observed agreement of MF, dyskinesia and FoG at baseline
The system detected MF in thirty-seven patients (95%), whilst dyskinesia and FoG were detected in fifteen (41%), and nine (23%), respectively.
However, the kappa coefficient for MF, dyskinesia and FoG was 0.089, 0.318 and 0.481 respectively (Table 3).
PD groups according to treatment adjustments
The treatment adjustments group included 29 patients was (85%) while there were 5 patients (15%) in the PD group without treatment adjustments. These 5 patients denied a modification of the treatment scheduled. In fact, we usually face this scenario in clinical practice when dealing with PD patients without having sensor information. Sometimes the patient doesn’t want to change the treatment because of their own decision because the intensity of Off or other PD symptoms are not so bothersome.
Levodopa was increased in 9 patients. In 2 patients the levodopa dose was decreased because adjustments of other dopaminergic drugs (opicapone and apomorphine infusion) were needed. Changes in levodopa schedule were indicated in 3 (10%) patients while a new treatment was added in 15 (52%) patients: opicapone 4, amantadine 3, oral/transdermic dopamine agonist 4, IMAOB (rasagiline/safinamide) 2, apomorphine infusion 1, antidepressant 1.
Clinical and WIS data after treatment changes
In the group with changes in the treatment schedule (85%) the increase of the levodopa equivalent dose was 86.1 ± 69.7 mg while there was an improvement of the UPDRS-III On total score from 21.52 (+/- 9.82) to 16.69 (+/- 6.11). When comparing the mean percentages of Off-Time and On-Time between the two groups, there were statistically significant differences (Table 4). In the intervention group the percentage of Off-Time decreased from 28.30 ± 11.15 to 23.96 ± 13.37 while it increased in the group without treatment changes from 17.02 ± 11.64 to 23.07 (p = 0.007). Regarding the On-Time, it increased from 26.36 ± 10.83 to 29.22 ± 13.58 in the intervention group and decreased from 39.52 ± 14.56 to 29.82 ± 13.07 in the group without any intervention (p = 0.002). Differences were also observed between groups in terms of number of steps and GF with less decrease in the group in which treatment was changed (p = 0.008 and 0.004 respectively). Change in other sensor-based parameters (dyskinesia, FoG and minutes walking per day) was not significant.
When analysing the variables of On and Off sensor values in the “intervention group”, it was found that the mean percentage of Off-Time among the patients who decreased their Off-Time (79% of patients) was -7.54 ± 5.26. The mean percentage of On-Time among the patients that increased their On-Time (59% of patients) was 8.9 ± 6.46.
When analysing the subgroup in which levodopa was increased (n = 9) against the group without treatment changes (n = 5), the WIS detected more On-Time in the patients in whom levodopa was increased (2.1 ± 6.7 vs– 9.7 ± 4.3; p = 0.003). More minutes walked (-1.1 ± 17.1 vs -22.0 ± 18.2; p = 0.039) and a greater number of steps (-281.3 ± 1976.8 vs– 3119.8 ± 1609.8; p = 0.014) were also found in this levodopa group. The GF worsen in the group without treatment changes (-0.8 ± 0.4) and remained unchanged in the levodopa group (0.0 ± 0.3) (p = 0.004).
Discussion
New technologies and wearable sensors based on machine-learning approaches can help to capture the response to dopaminergic treatments and provide complementary information besides the clinical scales and patient’s subjective opinion about the treatment response. In the present study, the STAT-ONTM system can detect significant sensor-based changes in terms of a decrease in the percentage of Off-Time, and increase of On-Time, number steps and GF after three months of having modified the treatment schedule. Besides, we have obtained a value of 8.9 ± 6.46 for the percentage of On-Time and -7.54 ± 5.26 for the Off-Time percentage. These preliminary results will need further replication for future studies with this sensor when analysing the magnitude of effect of interventions (either pharmacological or non-pharmacological such as physiotherapy). We have also found that the number of steps increased significantly, yet the minutes walking per day did not, meaning that the walking speed increased. Hence, with the wearable sensors not only the Off-Time/On-Time is important to analyse, but also gait variables such as GF and gait speed.
Other studies with different wearable-based technology have tried to demonstrate data-sensor changes after dopaminergic treatments in real-world settings [20,21,24–26,28–30]. Wrist sensors can overestimate symptoms more than waist-worn sensors, particularly tremor and dyskinesia [53]. Nonetheless, if the objective is to monitor the change measured by the sensor rather than accurately record PD symptoms, this can be achieved with either wrist or waist sensors [53]. In a recent study were PD patients used a smartphone App while doing 5 activities (voice, finger tapping, gait, balance, and reaction time), a mobile PD score (mPDS) derived from a novel machine-learning approach, captured improvement in response to dopaminergic therapy [20]. The watch device Personal KinetiGraph (PKG) has been widely used in the setting of routine clinical practice to identify a clinically significant response to levodopa [21], assess the changes in the dopaminergic oral therapy [24,28,29], identify candidates for second line therapies [24,50], or determine the changes in motor fluctuations following deep brain stimulation [26].
Wearable technologies provide additional and useful objective data that can enhance an enrichment of the clinical decision-making process, empowering the patient in the treatment decisions [22,23]. In our study, the treatment decisions were done according to the best clinical practice guidelines, but we are aware that the STAT-ONTM information could have influenced these treatment decisions. However, in the present study the WIS report was available for all the patients in the second visit. Hence, we don’t have measured how the impact of the STAT-ONTM information can influence in the treatment decisions. Nonetheless, in the ongoing multicentre, randomised clinical trial MoMoPa-EC, in which the STAT-ONTM is used, this issue will be analysed because there are three study arms (sensor information vs Hauser diaries vs clinical information only collected in the visit) [35]. In fact, other studies using wearable technology demonstrate that more treatment plans are performed with the additional information captured by the sensors, probably improving the patient dialogue [24,25,28,29].
This device is only able to detect the Off-Time by analysing the bradykinetic gait while the patient is walking and wearing the sensor [36,37]. According to the previous studies of this sensor, the sensitivity and specificity of the On/Off STAT-ONTM algorithms are of 97% and 88% respectively [39,40,42]. According to our results, this wearable system detects Off state in almost all the patients (95%) at baseline. Still, there was no agreement between the clinical and the WIS results when applying the Kappa analysis. Nonetheless, these results could be influenced by the fact that we don’t have used the Hauser diaries for the correlation. In this line, in the previous studies for the elaboration of the STAT-ON sensor algorithms, patients that didn’t experience MF or were unable to recognize them, were excluded of the studies [42,47]. On the contrary, in the present pilot study we have included patients with initial MF or that they did not manifested WO. However, a recent sub-analysis of the MoMoPa-EC study that analyse the agreement between the Hauser diaries and the STAT-ON sensor, show an intraclass correlation coefficient (ICC) of both methods of 0.57 (95% CI: 0.3–0.73) for the Off-Time and 0.48 (95% CI: 0.17–0.68) for the On-Time [54].
The disagreement in the detection of MF between the clinical scale and the sensor, especially in the scenario of the daily-clinical practice, opens the discussion about whether the sensor is detecting the problem before the patient understands the reason for it. The transition from the best patient’s motor state to worst motor state can be gradual and ambiguous for some patients, leading to imprecise evaluation [4]. Besides, some barriers exist in the clinical practice when evaluating the off periods. These barriers could be lack of awareness of the symptom among patients, cognitive impairment, reluctance to discuss symptoms, caregiver absence, lack of time’s physician to evaluate all the symptoms or physician’s lack of appreciation of the Off periods [55]. In this regard, the use of wearable medical devices can be crucial and potentially help the clinicians in the detection of MF [56]. Future studies with larger samples using specific WO scales and exploring the ability of STAT-ONTM of capturing initial MF are mandatory.
With regard to dyskinesia, the sensitivity and specificity for strong or mild trunk dyskinesia are 95% and 93% respectively, and the sensitivity is lower (39%) for mild limb dyskinesias [44]. In the sub-analysis of the MoMoPa trial mentioned before, the ICC of the Hauser diaries and the STAT-ON was 0.48 (95% CI: 0.17–0.68) [54]. The STAT-ON cannot detect all the type of dyskinesia along all the time of monitoring. Besides, the sensor only detects dyskinesia when the patient is not walking. This means, when the bradykinesia algorithm is not working. The kappa agreement for dyskinesia in the present study is of 0.318. Disagreement in dyskinesia can be explained because either the patient may not be aware of their dyskinesia, or the patient is doing an activity that might imitate a “false dyskinesia” like dancing. In the present study we tried to control the patient’s activities in a simple diary. However, we couldn’t control at all the activities potentially responsible for “false positive dyskinesia” such as cleaning, sweeping or going by car or public transport. With respect to the FoG, in the previous algorithm validation study of this WIS, the sensitivity and specificity were around 85% [43]. To our knowledge, there are still no studies that have analysed the agreement between the clinical detection of FoG and the STAT-ONTM sensor detection of FoG in an unsupervised scenario. Although in our study there is the limitation that specific FoG detection clinical scales have not been used, a kappa coefficient of 0.481 have been obtained. Hence, it will be of interest to have future results analysing this issue.
The results of the Spearman correlation regarding the UPDRS-III-Factor I and the STAT-ONTM measures have been previously studied. In the study of Rodríguez-Molinero et al, the correlation between the algorithm outputs and the UPDRS-III- Factor I was -0.67 (p < 0.01) [46]. In this study the comparison was directly between the Factor I and the GF while in our study we have obtained the variation of the mean GF of the same patient in the two monitorizations, which has 0.6. Although these results cannot be directly compared, the correlation obtained in our study shows consistent results with the previous ones. Recently, Pérez-López and collaborators, have found a Spearman correlation of– 0.63 (p> 0.001) between the mean fluidity and the UPDRS-III- Factor I [54].
The present work has several limitations. Due to the design of the study, heterogeneity in both the PD sample and the treatment groups couldn’t have been avoided. Moreover, we haven’t used other specific clinical scales for evaluating the WO, dyskinesia or FoG, nor the Hauser diaries have been used. Finally, the interesting point of the study that is performed in a “free-living-environment” implies that some false positives are unavoidable.
Despite the limitations of our study, we think that this wearable can provide additional value to PD neurologists for a better understanding of the PD patient ‘s motor state and a guiding for treatment decisions in the setting of routine clinical practice [8,23,57,58].
In summary, it’s feasible to use this sensor for monitoring PD treatment under ambulatory conditions. This system could serve as a complementary tool to assess PD motor symptoms and motor complications, although more studies with larger and homogeneous PD samples should be completed.
References
- 1. Kalia LV, Lang AE. Parkinson’s disease. Lancet 2015; 29: 896–912. Epub 2015 Apr 19.
- 2. Stocchi F, Jenner P and Obeso JA. When do levodopa motor fluctuations first appear in Parkinson’s disease? Eur Neurol 2010;63(5):257–66. Epub 2010 Mar 24. pmid:20332641
- 3. Martínez-Fernández R, Schmitt E, Martínez-Martin P and krack P. The hidden sister of motor fluctuations in Parkinson’s disease: A review on nonmotor fluctuations. Mov Disord 2016 Aug;31(8):1080–94. Epub 2016 Jul 19. pmid:27431515
- 4. Antonini A, Martinez-Martin P, Chaudhuri RK, Marello M, Hauser R, Katzenschlager R, et al. Wearing-Off Scales in Parkinson’s Disease: Critique and Recommendations. Mov Disord 2011 Oct;26(12):2169–75. Epub 2011 Jul 20. pmid:21780180
- 5. Stacy M. The wearing-off phenomenon and the use of questionnaires to facilitate its recognition in Parkinson’s disease. J Neural Transm 2010 Jul;117(7):837–46. Epub 2010 Jun 20. pmid:20563826
- 6.
Fahn S.; Elton R.L. Recent developments in Parkinson’s disease. Vol 2. Macmillan Health Care information: Florham Park, NJ; 153–164 1987.
- 7. Goetz CG, Stebbins GT, Chmura TA, Fahn S, Poewe W and Tanner C. Teaching program for the Movement Disorder Society-Sponsored Revision of the Unified Parkinson’s Disease Rating Scale: (MDS-UPDRS). Movement Disorders 2010; Jul 15;25(9):1190–4. pmid:20629138
- 8. Monje M. H. G, Foffani G, Obeso J and Sánchez-ferro A. New Sensor and Wearable Technologies to Aid in the Diagnosis and Treatment Monitoring of Parkinson’s Disease. Annu Rev Biomed Eng 2019; Jun 4; 21:111–143. pmid:31167102
- 9. Hauser RA, Friedlander J, Zesiewicz TA, Adler CH, Seeberger LC, O’Brien CFO, et al. A home diary to assess functional status in patients with Parkinson’s disease with motor fluctuations and dyskinesia. Clin Neuropharmacol 2000; 23(2):75–81. pmid:10803796
- 10. Hauser RA, Deckers F, Lehert P. Parkinson’s disease home diary: further validation and implications for clinical trials. Mov Disord 2004; Dec;19(12):1409–13. pmid:15390057
- 11. Papapetropoulos S. Patient Diaries as a Clinical Endpoint in Parkinson’s Disease Clinical Trials. CNS Neurosci Ther 2012; May;18(5):380–7. Epub 2011 Jun 7. pmid:22070400
- 12. Del Din S, Godfrey A, Mazzà C, Lord S, Rochester L. Free-living monitoring of Parkinson’s Disease: Lessons from the Field. Mov Disord 2016; Sep;31(9):1293–313. Epub 2016 Jul 25. pmid:27452964
- 13. Thorp JE, Adamczyk PG, Ploeg HL and Pickett KA. Monitoring motor symptoms during activities of Daily Living Individuals with Parkinson’s Disease. Front Neurol 2018; Dec 12;9:1036. eCollection 2018
- 14. Rovini E, Maremmani C and Cavallo F. How wearable sensors can support Parkinson’s Disease diagnosis and treatment: a systematic review. Front Neurosci 2017 Oct 6; 11:555. pmid:29056899
- 15. Godinho C, Domingos J, Cunha G, Santos AT, Fernandes RM, Abreu D, et al. A systematic review of the characteristics and validity of monitoring technologies to assess Parkinson’s disease. Journal of Neuro Engineering and Rehabilitation 2016 Mar 12; 13:24. pmid:26969628
- 16. Artusi CA, Mishra M, Latimer P, Vizcarra JA, Lopiano L, Maetzler W, et al. Integration of technology-based outcome measures in clinical trials of Parkinson and other neurodegenerative diseases. Park Relat Disord 2018 Jan;46 Suppl 1(Suppl 1): S53–S56. Epub 2017 Jul 26. pmid:28760593
- 17. Maetzler W, Domingos J, Srulijes K, Ferreira JJ and Bloem BB. Quantitative wearable sensors for objective assessment of Parkinson’s disease. Mov Disord 2013 Oct;28(12):1628–37. Epub 2013 Sep 12. pmid:24030855
- 18. Sánchez-Ferro Á, Elshehabi M, Godinho C, Salkovic D, Hobert MA, Domingos J, et al. New methods for the assessment of Parkinson’s disease (2005 to 2015): a systematic review. Mov Disord 2016; Sep;31(9):1283–92. Epub 2016 Jul 19. pmid:27430969
- 19. Mancini M, Bloem BR, Horak FB, Lewis SJG, Nieuwboer , and Nonnekes J. Clinical and methodological challenges for assessing freezing of gait: future perspectives. Mov Disord 2019 Jun;34(6):783–790. Epub 2019 May 2. pmid:31046191
- 20. Zhan A, Mohan S, Tarolli C, Schneider RB, Adams JL, Sharma S, et al. Using smartphones and machine learning to quantify Parkinson Disease Severity. The mobile Parkinson disease score. JAMA Neurol 2018 Jul 1;75(7):876–880. pmid:29582075
- 21. Khodakarami HK, Ricciardi L, Contarino MF, Pahwa R, Lyons KE, Geraedts VJ, et al. Prediction of the levodopa challenge test in Parkinson’s disease using data from a wrist-worn sensor. Sensors 2019 Nov 25;19(23):5153. pmid:31775289
- 22. Ferreira JJ, Godinho C, Santos AT, Domingos J, Abreu D, Lobo R, et al. Quantitative home-based assessment of Parkinson’s disease: The SENSE-PARK feasibility and usability study. BMC Neurology 2015 Jun 10; 15:89. pmid:26059091
- 23. Van Uem JMT, Maier KS, Hucker S, Scheck O, Hobert MA, Santos TA, et al Twelve-week sensor assessment in Parkinson’s disease: Impact on quality of life. Mov Disord 2016 Sep;31(9):1337–8. Epub 2016 May 31. pmid:27241524
- 24. Farzanehfar P, Woodrow H, Braybrook M, McGregor S, Evans A, Nicklason F et al. Objective measurement in routine care of people with Parkinson’s disease improve outcomes. NPJ Parkinsons Dis 2018 Apr 3; 4:10. eCollection 2018
- 25. Isaacson SH, Boroojerdi B, Waln O, McGraw M, Kreitzman DL, Klos K, et al. Effect of using a wearable device on clinical decision-making and motor symptoms in patients with Parkinson’s disease starting transdermal rotigotine patch: a pilot study. Park Relat Disord 2019Jul; 64:132–137. Epub 2019 Jan 28. pmid:30948242
- 26. Horne MK, McGregor S and Bergquist F. An objective fluctuation score for Parkinson’s disease. PLOS One 2015; 10(4): e0124522. pmid:25928634
- 27. Ossig C, GAndor F, Fauser M, et al. Correlation of quantitative motor state assessment using a Kinetograph and patient diaries in advanced PD: data from an observational study. Ossig C, Gandor F, Fauser M, et al. PLOS One 2016; 11(8): e0161559. pmid:27556806
- 28. Joshi R, Bronstein JM, Keener A, Alcazar j, Yang DD, Joshi M, et al. PKG movement recording system use shows promise routine clinical care of patients with Parkinson’s disease. Front Neurol 2019 Oct 1; 10:1027. eCollection 2019
- 29. Santiago A, Langston JW, Gandhy R, Dhall R, Brillman S, Rees L, et al. Qualitative evaluation of the Personal KinetiGraph TM movement recording system in a Parkinson’s clinic. J Parkinson’s Disease 2019;9(1):207–219. pmid:30412506
- 30. Thomsen TH, Wesenberg Kjaer T, Bastrup Jorgensen L, Haahr A and Winge K. Does the response to morning medication predict ADL-level of the day in Parkinson’s disease? Parkinson’s disease 2020 Jul 27; 2020:7140984. eCollection 2020
- 31. Silva de Lima AL, Evers LJW, Hahn T, de Vries NM, Daeschler M, Boroojerdi B, et al et al. Impact of motor fluctuations on real-life gait in Parkinson’s disease. Gait Posture 2018 May; 62:388–394. pmid:29627498
- 32. Pulliam CL, Heldman DA, Brokaw EB, Mera TO, Mari ZK, Burack MA. Continuous assessment of levodopa response in Parkinson’s disease using wearable motion sensors. IEEE Trans Biomed Eng. (2018) 65:159–64. pmid:28459677
- 33. Luis-Martínez R, Monje MHG, Antonini A, SánchezFerro Á, Mestre TA. Technology-enabled care: integrating multidisciplinary care in Parkinson’s disease through digital technology. Front Neurol. (2020) 11:1–10. pmid:33250846
- 34. Santos-García D, Ariztegui López N, Cubo E, Vinagre Aragón A, García-ramos R, Borrué C, et al. Clinical utility of a personalized and long-term monitoring device for Parkinson’s disease in a Real Clinical Practice Setting: an expert Opinion survey on STAT-ONTM. Neurologia (Engl Ed). 2020 Dec 24; S0213-4853(20)30339-X. pmid:33358530
- 35. Rodríguez-Molinero A, Hernández-Vara J, Miñarro A, Pérez-López C, Bayés-Rusiñol À, et al. Multicentre, randomized, single-blind, parallel group trial to compare the effectiveness of a Holter for Parkinson’s symptoms against other clinical monitoring methods: study protocol. BMJ Open 2021 Jul 19;11(7): e045272. pmid:34281918
- 36. Demers L, Weiss-Lambrou R and Ska B. Development of the Quebec User Evaluation of Satisfaction with assistive Technology (QUEST). Assist Technol 1996; 8:3–13. pmid:10159726
- 37.
Brooke J. (1996). SUS: A quick and dirty usability scale. In Jordan P.W., Thomas B., Weerdmeester B. A. & McClelland I. L. (Eds.), Usability Evaluation in Industry (pp. 189–194). London: Taylor & Francis.
- 38. Caballol N, Prats A, Quispe P, Ranchal MA, Alcaine S, Fondevilla F, Bayés A. Early detection of Parkinson’s disease motor fluctuations with a wearable inertial sensor [abstract]. Mov Disord. 2020; 35 (suppl 1) S643–S644.
- 39. Pérez-López C, Samà A, Rodríguez-Martín D, Català A, Cabestany J, Moreno-Arostegui JM, et al. Assessing motor fluctuations in Parkinson’s disease patients based on a single inertial sensor. Sensors 2016; Dec 15;16(12):2132. pmid:27983675
- 40. Samà A, Pérez-López C, Rodríguez-Martín D, Català A, Moreno-Arostegui JM, Cabestany J, et al. Estimating bradykinesia severity in Parkinson’s disease by analysing gait through a waist-worn sensor. Comput Biol Med and Medicine 2017 May 1; 84:114–123. Epub 2017 Mar 23. pmid:28351715
- 41. Rodríguez-Martín D, Pérez-López C, Samà A, Català A, Moreno Arostegui JM, Cabestany J, et al. et al. A waist-worn inertial measurement unit for long-term monitoring of Parkinson’s disease patients. Sensors 2017 Apr 11;17(4):827. pmid:28398265
- 42. Bayés À, Samà A, Prats A, Pérez-López C, Crespo-Maraver M, Moreno JM, et al. A “HOLTER” for Parkinson’s disease: Validation of the ability to detect on-off states using the REMPARK system. Gait Posture 2018 Jan;59:1–6. Epub 2017 Sep 25. pmid:28963889
- 43. Rodríguez-Martín D, Samà A, Pérez-López C, Català J, Moreno-Arostegui JM, casbestany J, et al et al. Home detection of freezing of gait using support vector machines through a single waist-worn triaxial accelerometer. PLOS One 2017 Feb 15;12(2):e0171764. pmid:28199357
- 44. Pérez-López C, Samà A, Rodríguez-Martín D, Moreno-Aróstegui JM, Cabestany J, Bayés À, et al. Dopaminergic-induced dyskinesia assessment based on a single belt-worn accerelometer. Artif Intell Med 2016 Feb;67:47–56. Epub 2016 Jan 14. pmid:26831150
- 45. Samà A, Rodríguez-Martín D, Pérez-López C, Català A, Alcaine S, Mestre B, et al. Determining the optimal features in freezing of gait detection through a single waist accelerometer in home environments. Pattern Recognition Letters 2017; 105: 135–143.
- 46. Rodríguez-Molinero A, Samà A, Pérez-López C, Rodríguez-Martín D, Quinlan LR, Alcaine S, et al. Analysis of correlation between an accerelometer-based algorithm for detecting parkinsonian gait and UPDRS subscales. Front Neurol 2017 Sep 1;8:431. eCollection 2017
- 47. Rodríguez-Molinero A, Pérez-López C, Samà A, de Mingo E, Rodríguez-Martín D, Hernández-vara J, et al. A kinematic sensor and algorithm to detect motor fluctuations in Parkinson disease: validation study under real conditions of use. JMIR Tehabil Assist Technol 2018; 25:5(1): e8. pmid:29695377
- 48. Daniel SE Lees AJ. Parkinson’s Disease Society Brain Bank London: overview and research. J Neural Transm Suppl 1993;39:165–72. pmid:8360656
- 49. Stebbins GT, Goetz CG, Lang AE, Cubo E. Factor analysis of the motor section of the unified Parkinson’s disease rating scale during off-state. Mov Disord 1999 Jul;14(4):585–9.
- 50.
N. Caballol, C. Pérez-López DR. STAT-ON: Ejemplos Prácticos. https://www.sense4care.com/wp-content/uploads/2020/05/Guia-de-interpretación-de-los-reportes.pdf.
- 51.
Samà A, Perez-Lopez C, Romagosa J, et al. Dyskinesia and motor state detection in Parkinson’s Disease patients with a single movement sensor. In: Haemmerich D, Panescu D, Pearce J, et al., eds. Engineering in Medicine and Biology Society, 2012. 34th Annual International Conference of the IEEE, San Diego. Vol 2012.; 2012:1194–1197. https://doi.org/10.1109/EMBC.2012.6346150 pmid:23366111
- 52.
Pérez-López C, Samà A, Rodríguez-Martín D, et al. Monitoring Motor Fluctuations in Parkinson’s Disease Using a Waist-Worn Inertial Sensor. In: Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol 9094.; 2015:461–474. https://doi.org/10.1007/978-3-319-19258-1_38
- 53. Kim DW, Hassett LM, Nguy V, and Allen NE. A comparison of activity monitor data from devices worn on the wrist and the waist in people with Parkinson’s disease. Mov Disord Clin Pract 2019 Oct 18;6(8):693–699. pmid:31745480
- 54. Pérez-López C, Hernández-Vara J, Caballol N, Bayés À, Buongiorno MT, et al. Comprison of the results of a Parkinson’s Holter monitor with patient diaries, in real conditions of Use: A sub-analysis of the MoMoPa-EC Clinical Trial. Front Neurol 2022 https://doi.org/10.3389/fneur.2022.835249.
- 55. Armstrong MJ, Rastgardani T, Gagliardi AR and Marras C. Barriers and facilitators of communication about off periods in Parkinson’s disease: Qualitative analysis of patient, carepartner, and physician Interviews. PLOS One 2019 Apr 18;14(4):e0215384. pmid:30998707
- 56. Rodríguez-Martín D, Cabestany J, Pérez-López C, Pie M, Calvet J, et al. A New Paradigm in Parkinson’s Disease Evaluation with wearable medical devices: A review of STAT-ONTM. Front Neurol 2022 Jun 2;13:912343. eCollection 2022
- 57. Espay A, Bonato P, Nahab FB, Maetzler W, Dean JM, Klucken J, et al. Technology in Parkinson’s disease: challenges and opportunities. Mov Disord 2016 Sep;31(9):1272–82. Epub 2016 Apr 29. pmid:27125836
- 58. Khodakarami H, Farzanehfar and Horne M. The use of data from the Parkinson’s KinetiGraph to identify potential candidates for device assisted therapies. Sensors (Basel) 2019 May 15;19(10):2241. pmid:31096576