Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Feasibility of large-scale deployment of multiple wearable sensors in Parkinson's disease

Feasibility of large-scale deployment of multiple wearable sensors in Parkinson's disease of large-scale deployment of multiple wearable sensors in Parkinson's disease. PLoS ONE 12(12): Wearable devices can capture objective day-to-day data about Parkinson's Disease (PD). e0189161. https://doi.org/10.1371/journal. This study aims to assess the feasibility of implementing wearable technology to collect data pone.0189161 from multiple sensors during the daily lives of PD patients. The Parkinson@home study is Editor: Oscar Arias-Carrion, Hospital General Dr. an observational, two-cohort (North America, NAM; The Netherlands, NL) study. To recruit Manuel Gea Gonzalez, MEXICO participants, different strategies were used between sites. Main enrolment criteria were Received: August 23, 2017 self-reported diagnosis of PD, possession of a smartphone and age18 years. Participants used the Fox Wearable Companion app on a smartwatch and smartphone for a minimum of Accepted: November 20, 2017 6 weeks (NAM) or 13 weeks (NL). Sensor-derived measures estimated information about Published: December 20, 2017 movement. Additionally, medication intake and symptoms were collected via self-reports in Copyright:© 2017 Silva de Lima et al. This is an the app. A total of 953 participants were included (NL: 304, NAM: 649). Enrolment rate was open access article distributed under the terms of 88% in the NL (n = 304) and 51% (n = 649) in NAM. Overall, 84% (n = 805) of participants the Creative Commons Attribution License, which permits unrestricted use, distribution, and contributed sensor data. Participants were compliant for 68% (16.3 hours/participant/day) of reproduction in any medium, provided the original the study period in NL and for 62% (14.8 hours/participant/day) in NAM. Daily accelerometer author and source are credited. data collection decreased 23% in the NL after 13 weeks, and 27% in NAM after 6 weeks. Data Availability Statement: The dataset used in Data contribution was not affected by demographics, clinical characteristics or attitude this article is owned by the Michael J Fox towards technology, but was by the platform usability score in the NL (χ (2) = 32.014, Foundation. In the original study consent forms, participants consented to share the full, de- p<0.001), and self-reported depression in NAM (χ (2) = 6.397, p = .04). The Parkinson@- identified study dataset only with qualified home study shows that it is feasible to collect objective data using multiple wearable sensors researchers. All qualified researchers in possession in PD during daily life in a large cohort. of a research question who would like to have access to the dataset used in this article, may do so through the Michael J. Fox Foundation Website (https://www.michaeljfox.org/page.html?id=192). A section of the website is devoted to sharing datasets, and the data used to support the PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 1 / 15 The Parkinson@Home study conclusions of this article will be available there. Introduction Interested and qualified researchers will be Parkinson's Disease (PD) is a common neurodegenerative disease in which patients experience required to register/apply and sign an Data Use both motor and non-motor symptoms[1]. Treatment is primarily based on the management Agreement that includes publication and IP policies. After registration, all enquiries for data of symptoms by increasing dopamine levels through pharmacological therapy or surgery[2, 3]. access will be judged by the Michael J Fox Additionally, non-pharmacological therapies, such as physiotherapy, occupational therapy or Foundation data sharing committee. speech therapy, are available to support patients[4]. Funding: This work was gratefully supported by Although good results in the management of motor symptoms have been achieved, particu- the Michael J Fox Foundation (https://www. larly in the early stages of the disease[5, 6], two major problems hamper long-term treatment. michaeljfox.org/), Intel Corporation (Tel Aviv, Israel First, current pharmacological therapy is successful for a limited period. In the long term, Ðhttps://www.intel.com) [grant number most patients develop unmanageable motor complications that can lead to worsening of qual- 10231.01], the Stichting Parkinson Fonds (https:// ity of life[7]. Second, evaluation of day-to-day variations in PD symptoms is difficult when www.parkinsonfonds.nl/), The Netherlands Organisation for Health Research and Development relying solely upon periodic consultations by clinicians[8]. Therefore, more detailed, objective (https://www.zonmw.nl/en/), Philips Research and reliable measures during daily living could potentially improve the management of PD. (https://www.philips.com/aw/research/home.html), Wearable sensors have been used to assess PD-related symptoms continuously and longitu- UCB pharma (http://www.ucb.com/), and the dinally during daily living[9±12]. Wearables may provide greater insight into a patient's Movement Disorders Society (https://www. disease status, allowing patients to self-manage their symptoms and monitor medication movementdisorders.org/MDS.htm). The author responses[13±18]. Furthermore, wearable sensor data may improve our scientific understand- Ana LõÂgia Silva de Lima is supported by Coordenac Ëão de Aperfeic Ëoamento de Pessoal de ing of disease progression by showing changes in motor and non-motor symptoms over time, NõÂvel SuperiorÐCAPES (http://www.capes.gov.br/) furthering the development of digital biomarkers for disease progression[19]. [grant number 0428-140]. The Michael J Fox While the potential value of wearable sensors for disease management and research are Foundation was involved in the data collection, increasingly becoming clear, various critical aspects of feasibility remain to be determined. interpretation of the data in the NAM cohort and Only a few studies have rigorously investigated the feasibility and acceptability of using a wear- manuscript revision. The funder Intel Corporation provided support in the form of salaries for authors able platform comprising a smartphone in combination to a smartwatch. Moreover, these EC and MA, participated in study design, data prior findings remained limited by the small sample sizes (biggest sample thus far: 40 PD collection and writing the report. The founders patients) [9, 13, 17, 18]. Therefore, we aimed to investigate the feasibility of using a wearable Philips Research and UCB pharma provided platform in a much larger sample of PD patients, with a focus on recruitment success, attrition support in the form of salaries for authors [Philips rates, user compliance and system usability. Research: HB; UCB pharma: KC, BB and DT] and participated in writing the report, but did not have any additional role in the study design, data Methods collection and analysis or decision to publish. The Between August/2015 and November/2016, a total of 953 PD patients from two cohorts specific roles of these authors are articulated in the (n = 304 in The Netherlands (NL) and n = 649 PD in North America (Unites States and Can- ªAuthor contributionsº section. All other sponsors did not have any role in study design; in the adaÐNAM) participated in the Parkinson@Home feasibility study. To investigate the feasibil- collection, analysis and interpretation of the data; in ity of the technology in different contexts, both cohorts used the same wearable platform, but the writing of the report; and in the decision to had distinct strategies for recruitment, retention and study period. These topics are described submit the paper for publication. separately (overview in Table 1). Competing interests: Ana LõÂgia Silva de Lima is supported by CoordenacËão de Aperfeic Ëoamento de Study design and population Pessoal de NõÂvel SuperiorÐCAPES [grant number 0428-140]. Tim Hahn is supported by a research The NL cohort. The population and study design applied in the NL are described in detail grant provided by the Stichting Parkinson Fonds. elsewhere.[20] In short, participants were recruited from support groups, internet communi- Luc JW Evers is supported by a research grant ties and through physiotherapists specialized in treating PD patients. Enrolment criteria provided by UCB and Philips Research. Nienke M were: (1)30 years of age, (2) possession of a smartphone using an Android OS version 4.2 de Vries is supported by a research grant from The and (3) self-reported diagnosis of PD. No exclusion criteria were applied beyond enrolment Netherlands Organisation for Health Research and criteria. Development. Eli Cohen and Michal Afek are employed by Intel. This affiliation does not alter our All enrolled participants received a single medical examination, based on the ªParkinson's adherence to PLOS ONE policies on sharing data Progression Markers Initiativeº (PPMI)[21]. This included the full MDS-UPDRS[22], the and materials. Lauren Bataille and Margaret Montreal Cognitive Assessment (MoCA)[23], and the Modified Schwab and England Activi- Daeschler are supported by the Michael J. Fox ties of Daily Living Scale.[24] The medical examination was performed by specially trained Foundation. Babak Boroojerdi, Kasper Claes and physiotherapists who are members of ParkinsonNet[25], a Dutch network of health PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 2 / 15 The Parkinson@Home study Dolors Terricabras are employed by UCB. This professionals specialised in PD management. At the end of the 13-weeks study period, all affiliation does not alter our adherence to PLOS enrolled participants evaluated the usability of the system through the System Usability Scale ONE policies on sharing data and materials. Max A (SUS)[26, 27], and were enquired about ability to use a smartphone (see APPENDIX). Finally, Little received research funding support from the participants had the option to continue using the platform or return the Pebble smartwatch. Michael J Fox Foundation and UCB. Heribert The NAM cohort. Study recruitment for the NAM cohort was entirely virtual through Baldus is employed by Philips Research. This affiliation does not alter our adherence to PLOS direct emails to subjects participating in the ªFox Insight online studyº, Facebook advertise- ONE policies on sharing data and materials. ments to targeted populations, and advertisements on Fox Trial Finder, a clinical trial match- Bastiaan R Bloem has served as an editorial board ing tool for people with PD.[28] Additional to the NL, the following enrolment criteria were member of Movement Disorders, currently serves applied: (1)18 years of age and (2) participation in the Fox Insight Online Study[29] as an editorial board member of Physiotherapy (Table 1). Canada, is Associate Editor for the Journal of In order to enrol, interested participants had to first register in the Fox Insight study (if they Parkinson's disease, received honoraria from serving on the scientific advisory board for Danone had not done so already). Through Fox Insight, each participant completed online surveys and Zambon, has received fees for speaking at about demographics, medical history, cognition, physical activity, symptoms and PD related conferences from AbbVie and Teva, and received medications and surgeries. Once enrolled in the Fox Insight, participants were eligible to regis- research support from the Netherlands ter for the NAM cohort of the Parkinson@Home study on a separate webpage. These users Organization for Scientific Research, the Michael J completed an online enrolment form which was reviewed by the study team to determine eligi- Fox Foundation, the Prinses Beatrix Foundation, the Stichting Parkinson Fonds, the National Parkinson bility. All study registrants received an email confirming their eligibility or non-eligibility. After Foundation and the Parkinson Vereniging. Marjan J finishing the 6-weeks study period, participants had the option to continue using the platform. Faber received grant support from the Michael J Fox Foundation, the Stichting Parkinson Fonds, Philips Research, The Netherlands Organisation for Wearable platform Health Research and Development and Health The Intel Pharma Analytics Platform used has been described in detail elsewhere.[20, 30] Holland. Briefly, it consists of the Fox Wearable Companion app, used on both a smartwatch and Table 1. Study design and procedure overview at the two cohorts. The Netherlands North America Recruitment strategies Through Internet communities ✓ ✓ Through support groups ✓ - Through physiotherapists ✓ - Enrolment criteria  30 years old ✓ - Dutch resident ✓ - Smartphone using Android OS version 4.2 or higher ✓ ✓ Self-reported PD ✓ ✓ 18 years old - ✓ Registered for Fox Insight study - ✓ English-speaking Canadian or United States resident - ✓ Exclusion criteria None ✓ ✓ Consent process Informative email ✓ ✓ Online digital consent form ✓ ✓ Study kit Pebble smartwatch ✓ ✓ Installation guide ✓ ✓ User manuals ✓ ✓ Clinical evaluations Assessment by physical therapist ✓ - Fox Insight online self-assessment surveys - ✓ Study duration Minimum of 6 weeks - ✓ Minimum of 13 weeks ✓ - Instruction for device usage Minimum of 5 hours a day - ✓ 24 hours, 7 days a week ✓ - Support model Call-center during working hours ✓ ✓ Technical support calls for non-data contributors ✓ - Support emails for non-contributors - ✓ Usability questionnaire - ✓ - https://doi.org/10.1371/journal.pone.0189161.t001 PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 3 / 15 The Parkinson@Home study Fig 1. (a) Fox Wearable Companion app main screen; (b) Fox Wearable Companion app activity graph; (c) Fox Wearable Companion app movement during sleep graph; (d) Fox Wearable Companion app symptom self-reports. ªReprinted from [Intel and Michael J Fox Foundation] under a CC BY license, with permission from [INTEL ], original copyright [2017]. https://doi.org/10.1371/journal.pone.0189161.g001 smartphone, and a cloud environment. In this study, a Pebble smartwatch, was used together with the patients' Android phones. 50 Hz accelerometer data were collected continuously from the smartwatch and streamed to the smartphone. Sensor analysis algorithms are applied to the aggregated (30 second interval) smartwatch accelerometer data in the app to estimate outcomes (i.e. levels of activity, tremor and move- ment during sleep). These estimated quantities are transmitted, via Wi-Fi or mobile data, to a cloud environment. They are also presented to the user by graphs and summary reports within the app. Additionally, users are able to set medication reminders, report actual medication intake and rate their symptoms (e.g. tremor, dyskinesia, rigidity, bradykinesia) within the mobile app (Fig 1). Both estimated outcomes and patients reported outcomes (PROs) are stored in the cloud environment. Study procedures at both cohorts Participants from both cohorts provided electronic consent and received a research kit con- taining a Pebble smartwatch, an installation guide and user manuals. Next, participants installed the Fox Wearable Companion App on their devices and were asked to wear the smartwatch and keep their smartphone with them as much as possible on either a 24/7 basis for 13-weeks study period (NL) or for a minimum of 5 hours a day, 7 days a week, for a 6-weeks study period (NAM). Additionally, participants reported their medication intake (i.e medication name and doses) and PD symptom severity using the app. A helpline was available during the study period for technical support. Support calls or emails were sent to participants from whom data were not collected for more than seven consecutive days. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 4 / 15 The Parkinson@Home study Outcome definitions and statistical analysis Feasibility assessment included recruitment, attrition, compliance and system usability. Recruitment success was analysed by (1) the total number of enrolled participants and (2) the number of eligible registrants that did not complete the informed consent. Compliance, simi- lar to previous studies[31, 32], was calculated as the median percentage of the study period where accelerometer data were collected. Attrition rate, based upon Eysenbach et al. [33], were measured by (1) decrease in the daily percentage of collected accelerometer data during each study period and (2) decrease in the number of participants contributing accelerometer data. Finally, system usability was measured by the median total score on the System Usability Scale. We investigated the relationship between self-reported demographics, clinical data, ability to use a smartphone (S1 File), System Usability score and the percentage of accelerometer data collected to identify factors that influence compliance levels. Participant demographic and clinical characteristics were grouped into categories either following previously described liter- ature (presence of depression[34]; presence of cognitive impairment[23]) or by convenience (age; educational level: a measure of the last completed level of education where low education was equal to high school or lower levels, middle education was equal to bachelor, and high education was equal to master or higher levels; Hoehn & Yahr stage and Modified Schwab and England scale). Because compliance was not normally distributed, the median and quartiles were used to divide participants into three compliance groups (low, middle and high). The first quartile was the cut-off for the low compliant group and third quartile for the high com- pliant group. Depending on the distribution of other variables in the analysis, either Chi- square, Fisher's Exact Test or Kruskal-Wallis were used to investigate significant differences between compliance groups considering demographics, clinical characteristics, ability to use a smartphone and System Usability score. Ethics standards This study was conducted in compliance with the Ethical Principles for Medical Research Involving Human Subjects, as defined in the Declaration of Helsinki. The study protocol and communication materials were approved by the local ethics committee (NL: CMO Arnhem- Nijmegen; NL53034.091.15; NAM: New England IRB: 15±046). Results Recruitment and sample characteristics In the NL cohort, 347 eligible PD patients were invited to participate. Among those invited, 43 refused to participate. The main refusal reasons were ªStudy protocol seems too burdensomeº (44%, n = 19), followed by ªPersonal circumstancesº (33%, n = 14). A total of 304 patients (enrolment rate = 88%) were enrolled. In the NAM cohort, from the 866 participants of the Fox Insight study who received a direct invitation to participate, 306 were enrolled (6% were ineligible). 344 additional participants were included from the remaining recruitment channels, with varied ineligibility rates. A total of 649 registrants (enrolment rate = 51%) were enrolled. In both cohorts, 953 participants were enrolled. From them, 805 were data contributors (participants that contributed at least one accelerometer data point during study period). Anal- ysis of the demographic characteristics of both cohorts showed that, in comparison to NA, the NL cohort presented more men (χ2 (1) = 9.5146, p<0.01); older (χ2 (2) = 16.435, p = 0.001) and higher educated (χ2 (2) = 25.270, p<0.001) PD included participants. The characteristics of all participants are presented in Table 2. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 5 / 15 The Parkinson@Home study Table 2. Demographic and disease related characteristics of participants. The Netherlands North America Total Data Non- p- Data Non- p- Data Non- contributors compliant value contributors compliant value contributors compliant (n = 291)* (n = 13)* (n = 514)* (n = 135)* (n = 805)* (n = 148)* Sex Men 168 (65%) - NA 260 (53%) 25 (51%) .763 454 (58%) 25 (51%) Age 50 29 (10%) 1 (8%) .523 101 (21%) 11 (22%) .676 130 (17%) 12 (19%) 51±69 207 (72%) 8 (61%) 310 (63%) 28 (57%) 517 (66%) 36 (58%) 70 53 (18%) 4 (31%) 81 (16%) 10 (20%) 134 (17%) 14 (23%) Education Level Low 51 (20%) NA NA 168 (35%) 21 (44%) .477 219 (30%) 21 (44%) Middle 103 (40%) NA NA 187 (39%) 16 (33%) 290 (39%) 16 (33%) High 101 (40%) NA NA 126 (26%) 11 (22%) 227 (31%) 11 (22%) Disease severity 0/1 73 (30%) 1 (100%) NA 235 (49%) 21 (47%) .580 308 (43%) 22 (48%) 2 127 (53%) 0 134 (28%) 10 (22%) 261 (36%) 10 (22%) 3 34 (14%) 0 98 (20%) 13 (29%) 132 (18%) 13 (29%) 4/5 6 (2%) 0 13 (3%) 1 (2%) 19 (3%) 1 (2%) Depression No 238 (97%) - NA 346 (70%) 26 (53%) .013 584 (79%) 26 (53%) Cognitive impairment No (>26) 124 (53%) 1 (100%) NA NA NA NA 124 (53%) 1 (100%) Independency level 70 36 (15%) 0 NA NA NA NA 36 (15%) 0 71±80 51 (22%) 0 NA NA 51 (22%) 0 81±90 109 (46%) 1 (100%) NA NA 109 (46%) 1 (100%) 91 41 (17%) 0 NA NA 41 (17%) 0 How easy is for you to Very easy 59(22%) 0 .211 NA NA NA 59(22%) 0 use a smartphone? Easy 117(44%) 3(50%) NA NA 117(44%) 3(50%) Neither easy 64(22%) 1 (17%) NA NA 64(22%) 1 (17%) nor difficult Difficult 20(7%) 2 (33%) NA NA 20(7%) 2 (33%) Very difficult 6(2%) NA NA 6(2%) MDS-UPDRS (Median) 52.5 (QR 35± 22 - NA NA 52.5 (QR 35± 22 69) 69) *Number of missing values differed across variables; only valid percentages are reported. SES: Socioeconomic status; Disease severity: assessed with Hoehn and Yarh stages at the NL cohort and estimated from self-reported at NAM cohort; MDS-UPDRS: Movement Disorders SocietyÐUnified Parkinson's Disease Rating Scale. -Depression was assessed with the Geriatric Depression Scale at NL and self-reported on NAM site; -Cognitive impairment was assessed using the Montreal Cognitive Assessment, -Independency level was measured by Modified Schwab and England Activities of Daily Living Scale. st rd NA = not assessed. QR: 1 and 3 quartiles. https://doi.org/10.1371/journal.pone.0189161.t002 Technical support to participants In both cohorts, the helpdesk consisted of two research assistants, available for 20 hours (NL) and 40 hours (NAM) per week. The actual workload was dependent on: (1) the number of participants simultaneously enrolled in the trial; and (2) the occurrence of bugs in the app or server downtime. The most frequent and time-consuming problems were: (1) Bluetooth dis- connection between the smartwatch and the smartphone and (2) questions regarding the med- ication report, especially in the first weeks of participation. Compliance Among both cohorts, 85% (n = 805 of 953 enrolled) of participants were data contributors. In the NL, 291 data-contributors collected data for a median of 1,478 hours each in the 13-weeks, PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 6 / 15 The Parkinson@Home study Fig 2. Distribution of compliance among all enrolled participants in the NL (n = 304-black) and NAM (n = 649-white) study cohorts. https://doi.org/10.1371/journal.pone.0189161.g002 st rd with quartile ranges (1 and 3 QR) of 888 to 1,827 hours. In NAM, 514 data contributors st rd collected a median of 621 hours (1 QR: 286 and 3 QR: 828 hours) each during the 6-weeks. st rd Compliance rates for each cohort were 68% (1 and 3 QR: 41%-83%) equal to 16.3 hours/ st rd participant/day in the NL and 62% (1 and 3 QR: 28%-82%) equal to 14.8 hours/participant/ day in NAM (Fig 2). Attrition In the NL, 13 participants (4% of all NL enrolled participants) did not contribute any data dur- ing the study period and were thus non-compliant. In NAM, this number was 135 (21% of all NAM enrolled participants). Additionally, 82 (27% of all enrolled) data-contributors in the NL became non-compliant during the study period. The primary known reasons (n = 47) were ªPersonal circumstancesº (38%, n = 18) and ªSystem too complex/System related issuesº (34%, n = 16). For the NAM cohort, although reasons were unknown, this number was 89 (17% of all enrolled). The attrition in the median percentage of sensor data collected daily varied between cohorts. In the NL, the attrition rate was 23% after 13-weeks' study period. In the NAM, attri- tion was 27% after 6-weeks' study period (Fig 3). Attrition in participation was tracked during and beyond the compulsory study period for each cohort. he number of participants decreased rapidly after the end of the study period in the NL cohort. A more gradual attrition in participation occurred in the NAM cohort (Fig 4). Ninety-six percent (n = 280) of data-contributors in the NL reported their medication through the app, while 78% (n = 404) did so in NAM. On average, data-contributors who used medication reports reported 351±217 medication intakes during the 13-week study period in PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 7 / 15 The Parkinson@Home study Fig 3. Attrition in compliance per day for NL (n = 291, black) and NAM participants (n = 514, gray) during the follow up period. https://doi.org/10.1371/journal.pone.0189161.g003 the NL and 127±113 over 6-week study period in NAM. Both cohorts showed a low and non- exponential attrition in medication report, similar to the attrition showed in compliance with the accelerometer data (data not shown). System usability In the NL cohort, 256 participants completed the System Usability Scale (response rate = 71.4%). st rd The median score was 62.5 (1 and 3 QT 47.5±72.5), which classifies the wearable platform in a category between ªOkº and ªGoodº (Fig 5). Factors related to compliance After grouping all NL data-contributors into compliance groups, analysis reveals no significant differences in the distribution of demographics, clinical characteristics and ability to use a smartphone between these groups. However, Kruskal-Wallis analysis demonstrates that the System Usability score reported is significantly different between the groups (χ (2) = 32.014, p<0.001). The mean rank score is 84.8 for the low compliant group, 130.8 for the middle com- pliant group and 160.0 for the high compliant group, which indicates that participants in the high compliant group provided a higher usability score to the system. For the NAM cohort, analysis shows that demographics and clinical characteristics between the three compliance groups was comparable, except for a trend regarding self-reported depression (χ (2) = 6.397, p = .04). This result indicates that a slightly higher number of self- reported depressed patients are in the low compliant group (Table 3). PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 8 / 15 The Parkinson@Home study End of follow-up End of follow-up Fig 4. Number of participants actively collecting sensor data at the NL (gray) and NAM (black) cohorts during and after the follow- up period (total initial n = 805). https://doi.org/10.1371/journal.pone.0189161.g004 Discussion This study assessed the feasibility of using a wearable platform for long-term data collection in a large sample of PD patients. We focused on: recruitment success, attrition rates, compliance Fig 5. SUS scoring of the Fox Wearable Companion platform (smartwatch with smartphone app) as rated by participants. https://doi.org/10.1371/journal.pone.0189161.g005 PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 9 / 15 The Parkinson@Home study Table 3. Distribution of data-contributors' characteristics and influence on compliance for the NL and NAM cohorts. The Netherlands North America Low Middle High p- Low Middle High p- compliance compliance compliance value compliance compliance compliance value (n = 73)* (n = 146)* (n = 72)* (n = 128)* (n = 256)* (n = 129)* 4 4 Sex Men 39 (71%) 84 (64%) 44 (65%) .63 64 (56%) 133 (54%) 62 (49%) .50 4 4 Age 50 7 (10%) 15 (10%) 7 (10%) .95 30 (26%) 49 (20%) 22 (17%) .41 51±69 51 (70%) 103 (71%) 53 (75%) 67 (57%) 161 (65%) 81 (64%) 70 15 (20%) 27 (19%) 11 (16%) 20 (17%) 36 (15%) 24 (19%) 4 4 Education Level Low 11 (20%) 23 (17%) 17 (25%) .68 46 (41%) 84 (35%) 38 (31%) .14 Middle 21 (38%) 45 (84%) 24 (35%) 43 (38%) 86 (35%) 57 (46%) High 23 (42%) 51 (39%) 27 (40%) 24 (21%) 73 (30%) 29 (23%) Disease severity 0+1 15 (42%) 37 (28%) 21 (30%) .55 50 (44%) 121 (50%) 64 (52%) .45 2 15 (42%) 72 (54%) 40 (57%) 30 (26%) 66 (27%) 38 (31%) 3 6 (16%) 20 (15%) 8 (11%) 30 (26%) 48 (19%) 19 (15%) 4+5 0 (0%) 5 (4%) 1 (1%) 4 (4%) 7 (3%) 2 (2%) 1 5 4 Depression No (<6) 36(100%) 135 (98%) 67 (94%) .28 74 (63.2%) 172 (70%) 99 (78%) .04 2 4 Cognitive impairment No (>26) 16 (47%) 74 (56%) 34 (50%) .58 - - - - Level of 70 5 (14%) 21 (16%) 10 (15%) .16 - - - - independency 71±80 13 (36%) 30 (23%) 8 (12%) - - - 81±90 13 (36%) 61 (46%) 35 (52%) - - - 91 5 (14%) 21 (16%) 15 (22%) - - - How easy is for you to Very easy 12 (19%) 33 (24%) 14 (21%) .06 - - - - use a smartphone? Easy 27 (44%) 60 (44%) 30 (44%) - - - Neither easy 14 (23%) 27 (20%) 23 (34%) - - - nor difficult Difficult 5 (8%) 14 (10%) 1 (2%) - - Very difficult 4 (6%) 2 (2%) (0%) - - - - MDS-UPDRS 49 55 48 .55 - - - - (median) SUS (median) 50 65 70 <0.001 - - - - *Missing values varied across variables. Red is significant at .05; green is significant at .001. SES: Socioeconomic status; Disease severity: assessed with Hoehn and Yarh stages at the NL cohort and estimated from self-reported at NAM cohort; GDS: Geriatric Depression Scale; MoCA: Montreal Cognitive Assessment, MDS-UPDRS: Movement Disorders SocietyÐUnified Parkinson's Disease Rating Scale, SUS: System Usability Scale. -Depression was assessed with the Geriatric Depression Scale at NL and self-reported on NAM; -Cognitive impairment was assessed using the Montreal Cognitive Assessment; -Independency level was measured by Modified Schwab and England Activities of Daily Living Scale; ±Pearson Chi-Square; ±Fisher's Exact Test; -Kruskal-Wallis https://doi.org/10.1371/journal.pone.0189161.t003 and system usability. Enrolment rate was 88% (n = 304) in the NL and 51% (n = 649) in NAM. Nearly 85% of all enrolled participants contributed sensor data during the study period. Median compliance rate was 68% (16.3 hours/participant/day) in the NL, and 62% (14.8 hours/participant/day) in NAM. The rate of accelerometer data collected each day declined 23% in the NL after 13-weeks of study period, and 27% in NAM after 6-weeks of study period. The distribution of demographics, clinical characteristics and ability to use a smartphone did not differ across compliance groups in the NL, but System Usability score did differ. For the PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 10 / 15 The Parkinson@Home study NAM, the distribution of demographics and clinical characteristics between the compliance groups was comparable, except for self-reported depression status. The high compliance in this study shows that it is feasible for people with PD to use this wearable platform in a real-world environment for many months. Although the feasibility of using consumer wearable sensors to monitor PD symptoms has been previous reported[9, 13, 17, 18, 31, 35], this is the first rigorous observational study to investigate the feasibility of a wear- able platform comprising a smartwatch combined with a smartphone in such a large patient group (the largest prior study included only 40 patients). Additionally, the small differences in study protocols across cohorts allowed us to observe the impact of varying usage instructions on compliance. Comparing the feasibility results obtained in this study to other studies, where either mobile apps were used in large cohorts[36] or e-health technologies were use[37, 38] d, we achieved a high compliance together with small and non-exponentially decreasing attrition rate, even though exponential decrease in compliance is the norm in these sort of studies[33]. This unusually high compliance rate may be attributed to the ªpassive” data collection. In this case, little or no interaction with the technology is required in order to collect sensor data. Participants using the Parkinson@Home wearable platform, other than reporting their medi- cation intake when reminded by the alarm (which was widely perceived as a service, instead of a burden), did not need to interact actively with the smartphone or smartwatch. In another similar smartphone-based study where ªactive”, ªtask-basedº monitoring was used (that is, where participants needed to perform certain specific tasks, at regular intervals prompted by the platform)[36], a more typical high and exponential attrition rate was observed. While it is difficult to draw firm conclusions from this comparison (because the two platforms are some- what different), we suspect that periodic and long interaction by users may increase attrition, leading to attrition rates seen in paper-based diaries[39]. The low and non-exponential attri- tion seen in the medication reports, a quick and less burdensome task, strengthened this conclusion. Thus, passive monitoring, where little to no interaction with the technology is required, may lead to better overall compliance rates. Despite the potential influence of age, gender and PD-related impairment (i.e. physical or cognitive) on compliance, our results showed that overall disease severity, MDS-UPDRS scores, independency level or cognitive impairment, did not influence compliance, which sug- gests that this platform could be used by most PD patients. The unique design of the Parkin- son@home study can partially explain this result. The presence of a personalized support centre, which was previously described as an effective strategy to improve retention of partici- pants[33], may have increased patients' confidence in using the system and have compensated for any disease-related difficulties. Moreover, the ªpro-activeº support model, with scheduled calls to participants who showed signs of low compliance, may have boosted compliance by providing a quick resolution of technical interruptions, and addressed any apathy towards participation caused by technology difficulties. This support is even more important because compliance is compromised in participants that reported low System Usability scores. Therefore, in order to achieve high compliance while using smartphone/smartwatch wearable platforms to measure PD related symptoms at home, it is beneficial to: (1) improve the platform's usability, (2) reduce the num- ber of technical issues, and (3) run a personalized support centre that can provide guidance to deal with possible technology related issues that participants may encounter. Limitations The Parkinson@Home study did have a few limitations. First, this is one of the first large-scale cohort studies using consumer wearable sensors in PD, with a long study period duration (i.e. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 11 / 15 The Parkinson@Home study up to 13 weeks). However, the study sample consisted only of PD patients that possessed a smartphone, thus introducing a possible selection bias, e.g. towards more highly educated sub- jects. Although smartphone penetration in the NL and NAM is high[40, 41], participants may not reflect the majority of PD patients living in the Netherlands, North America, or elsewhere in the world. Furthermore, when compared to the general PD population[42], participants were mainly young with a mild disease impairment and with some degree of cognitive impaired. Even though these variables showed no obvious influence on compliance, a more impaired population may need more personal support in order to maintain compliance. Future studies should aim for a more stratified population in order to further confirm the lack of influence across the full range of disease severity in the compliance with wearable sensors among PD patients. Second, the present results only apply to the use of two specific consumer grade devices (i.e. smartphone and smartwatch). Although consumer grade devices bring potential advantage over the use of dedicated medical devices, it is unknown whether our promising feasibility results would generalize to dedicated medical devices which are often more expensive and less user-friendly. Conclusion In conclusion, the Parkinson@home trial showed that it is feasible to deploy a technology plat- form consisting of consumer-grade wearable and mobile devices for long-term data collection in a large and geographically diverse PD population. Importantly, compliance was comparable for patients with a range of backgrounds, including men and women, different ages, and some variations in disease severity. These findings suggest that wearables may offer a promising approach to overcome the limitations in monitoring disease status and progression of mildly impaired PD patients in a real-life environment. The platform here used is a promising and practical approach to capturing large amounts of sensor data from many participants by pas- sive means, without much need for interaction with the technology. In the future, these prop- erties may position sensor technologies as effective tools for monitoring PD and the ªlived experienceº of PD patients. Supporting information S1 File. Scale created by the researchers to measure ability to use a smartphone. (TIF) Acknowledgments We thank the many people with Parkinson's disease who participated in these studies and the Stichting Parkinson Nederland. Without them this work would not have been possible. Author Contributions Conceptualization: Ana Lõgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Eli Cohen, Michal Afek, Lauren Bataille, Margaret Daeschler, Max A. Little, Bastiaan R. Bloem, Marjan J. Faber. Data curation: Ana Lõgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Nienke M. de Vries, Eli Cohen, Michal Afek, Lauren Bataille, Margaret Daeschler. Formal analysis: Ana Lõ Âgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Lauren Bataille, Marga- ret Daeschler, Marjan J. Faber. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 12 / 15 The Parkinson@Home study Funding acquisition: Bastiaan R. Bloem, Marjan J. Faber. Investigation: Ana Lõ Âgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Nienke M. de Vries, Lau- ren Bataille, Margaret Daeschler, Bastiaan R. Bloem, Marjan J. Faber. Methodology: Bastiaan R. Bloem, Marjan J. Faber. Project administration: Bastiaan R. Bloem, Marjan J. Faber. Resources: Lauren Bataille, Margaret Daeschler. Software: Eli Cohen, Michal Afek. Supervision: Bastiaan R. Bloem, Marjan J. Faber. Visualization: Ana Lõ Âgia Silva de Lima, Tim Hahn. Writing ± original draft: Ana Lõ Âgia Silva de Lima, Tim Hahn, Luc J. W. Evers. Writing ± review & editing: Ana Lõ Âgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Nienke M. de Vries, Eli Cohen, Michal Afek, Lauren Bataille, Margaret Daeschler, Kasper Claes, Babak Boroojerdi, Dolors Terricabras, Max A. Little, Heribert Baldus, Bastiaan R. Bloem, Marjan J. Faber. References 1. Gelb DJ, Oliver E, Gilman S. Diagnostic criteria for Parkinson disease. Archives of neurology. 1999; 56 (1):33±9. PMID: 9923759 2. Goetz CG, Poewe W, Rascol O, Sampaio C. Evidence-based medical review update: Pharmacological and surgical treatments of Parkinson's disease: 2001 to 2004. Movement Disorders. 2005; 20(5):523± 39. https://doi.org/10.1002/mds.20464 PMID: 15818599 3. Mehanna R, Lai EC. Deep brain stimulation in Parkinson's disease. Translational neurodegeneration. 2013; 2(1):22. https://doi.org/10.1186/2047-9158-2-22 PMID: 24245947 4. Cutson TM, Laub KC, Schenkman M. Pharmacological and Nonpharmacological Interventions in the Treatment of Parkinson's Disease. Physical Therapy. 1995; 75(5):363±73. PMID: 7732080 5. Lewis S, Foltynie T, Blackwell A, Robbins T, Owen A, Barker R. Heterogeneity of Parkinson's disease in the early clinical stages using a data driven approach. Journal of Neurology, Neurosurgery & Psychia- try. 2005; 76(3):343±8. 6. Giugni JC, Okun MS. Treatment of advanced Parkinson's disease. Current opinion in neurology. 2014; 27(4):450±60. Epub 2014/07/01. https://doi.org/10.1097/WCO.0000000000000118 PMID: 24978634. 7. Keus SHJ, Munneke M, Graziano M, Jaana P. European Physiotherapy Guideline for Parkinson's dis- ease. KNGF/ParkinsonNet, the Netherlands. 2014. 8. Mera TO, Heldman DA, Espay AJ, Payne M, Giuffrida JP. Feasibility of home-based automated Parkin- son's disease motor assessment. Journal of neuroscience methods. 2012; 203(1):152±6. https://doi. org/10.1016/j.jneumeth.2011.09.019 PMID: 21978487 9. Arora S, Venkataraman V, Zhan A, Donohue S, Biglan K, Dorsey E, et al. Detecting and monitoring the symptoms of Parkinson's disease using smartphones: a pilot study. Parkinsonism & related disorders. 2015; 21(6):650±3. 10. Patel S, Chen BR, Mancinelli C, Paganoni S, Shih L, Welsh M, et al. Longitudinal monitoring of patients with Parkinson's disease via wearable sensor technology in the home setting. Conf Proc IEEE Eng Med Biol Soc. 2011; 2011:1552±5. Epub 2012/01/19. https://doi.org/10.1109/IEMBS.2011.6090452 PMID: 11. Maetzler W, Domingos J, Srulijes K, Ferreira JJ, Bloem BR. Quantitative wearable sensors for objective assessment of Parkinson's disease. Mov Disord. 2013; 28(12):1628±37. https://doi.org/10.1002/mds. 25628 PMID: 24030855. 12. Hobert MA, Maetzler W, Aminian K, Chiari L. Technical and clinical view on ambulatory assessment in Parkinson's disease. Acta Neurol Scand. 2014; 130(3):139±47. https://doi.org/10.1111/ane.12248 PMID: 24689772. 13. Tsanas A, Little MA, McSharry PE, Ramig L. Using the cellular mobile telephone network to remotely monitor parkinsons disease symptom severity. IEEE Transactions on Biomedical Engineering. 2012. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 13 / 15 The Parkinson@Home study 14. Patel S, Lorincz K, Hughes R, Huggins N, Growdon J, Standaert D, et al. Monitoring motor fluctuations in patients with Parkinson's disease using wearable sensors. IEEE transactions on information technol- ogy in biomedicine: a publication of the IEEE Engineering in Medicine and Biology Society. 2009; 13 (6):864±73. Epub 2009/10/23. 15. Griffiths RI, Kotschet K, Arfon S, Xu ZM, Johnson W, Drago J, et al. Automated assessment of bradyki- nesia and dyskinesia in Parkinson's disease. Journal of Parkinson's disease. 2012; 2(1):47±55. https:// doi.org/10.3233/JPD-2012-11071 PMID: 23939408 16. Tzallas AT, Tsipouras MG, Rigas G, Tsalikakis DG, Karvounis EC, Chondrogiorgi M, et al. PERFORM: A System for Monitoring, Assessment and Management of Patients with Parkinson's Disease. Sensors. 2014; 14(11):21329±57. https://doi.org/10.3390/s141121329 PMID: 25393786 17. Lakshminarayana R, Wang D, Burn D, Chaudhuri KR, Cummins G, Galtrey C, et al. Smartphone-and internet-assisted self-management and adherence tools to manage Parkinson's disease (SMART-PD): study protocol for a randomised controlled trial (v7; 15 August 2014). Trials. 2014; 15(1):374. 18. Sharma V, Mankodiya K, De La Torre F, Zhang A, Ryan N, Ton TG, et al., editors. SPARK: personal- ized parkinson disease interventions through synergy between a smartphone and a smartwatch. Inter- national Conference of Design, User Experience, and Usability; 2014: Springer. 19. Espay AJ, Bonato P, Nahab FB, Maetzler W, Dean JM, Klucken J, et al. Technology in Parkinson's dis- ease: Challenges and opportunities. Movement Disorders. 2016; 31(9):1272±82. https://doi.org/10. 1002/mds.26642 PMID: 27125836 20. de Lima ALS, Hahn T, de Vries NM, Cohen E, Bataille L, Little MA, et al. Large-Scale Wearable Sensor Deployment in Parkinson's Patients: The Parkinson@ Home Study Protocol. JMIR Research Protocols. 2016; 5(3). 21. Marek K, Jennings D, Lasch S, Siderowf A, Tanner C, Simuni T, et al. The parkinson progression marker initiative (PPMI). Progress in neurobiology. 2011; 95(4):629±35. https://doi.org/10.1016/j. pneurobio.2011.09.005 PMID: 21930184 22. Goetz CG, Tilley BC, Shaftman SR, Stebbins GT, Fahn S, Martinez-Martin P, et al. Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): Scale pre- sentation and clinimetric testing results. Movement disorders. 2008; 23(15):2129±70. https://doi.org/10. 1002/mds.22340 PMID: 19025984 23. Hoops S, Nazem S, Siderowf A, Duda J, Xie S, Stern M, et al. Validity of the MoCA and MMSE in the detection of MCI and dementia in Parkinson disease. Neurology. 2009; 73(21):1738±45. https://doi.org/ 10.1212/WNL.0b013e3181c34b47 PMID: 19933974 24. Schwab R, England A. Projection Technique for Evaluating Surgery in Parkinson's Disease: Third Sym- posium on Parkinson's Disease. Edinburgh: Gillingham and Donaldson. 1969. 25. Bloem BR, Munneke M. Revolutionising management of chronic disease: the ParkinsonNet approach. BMJ. 2014; 348:g1838. https://doi.org/10.1136/bmj.g1838 PMID: 24647365 26. Bangor A, Kortum P, Miller J. Determining what individual SUS scores mean: Adding an adjective rating scale. Journal of usability studies. 2009; 4(3):114±23. 27. Brooke J. SUS-A quick and dirty usability scale. Usability evaluation in industry. 1996; 189(194):4±7. 28. Meunier CC, Chowdhury S, Sherer T, Brooks DW. Innovative Web-based Matching Service, Fox Trial Finder, as a mechanism to improve clinical trial recruitment. Movement Disorders. 2012; 27(4):e2. 29. Heusinkveld L, Hacker M, Turchan M, Bollig M, Tamargo C, Fisher W, et al. Patient Perspectives on Deep Brain Stimulation Clinical Research in Early Stage Parkinson's Disease. Journal of Parkinson's disease. 2017; 7(1):89±94. https://doi.org/10.3233/JPD-161031 PMID: 27911344 30. Cohen S, Bataille LR, Martig AK. Enabling breakthroughs in Parkinson's disease with wearable technol- ogies and big data analytics. mHealth. 2016; 2:20. Epub 2017/03/16. https://doi.org/10.21037/mhealth. 2016.04.02 PMID: 28293596. 31. Ferreira JJ, Godinho C, Santos AT, Domingos J, Abreu D, Lobo R, et al. Quantitative home-based assessment of Parkinson's symptoms: The SENSE-PARK feasibility and usability study. BMC neurol- ogy. 2015; 15(1):89. 32. Brannon EE, Cushing CC, Crick CJ, Mitchell TB. The promise of wearable sensors and ecological momentary assessment measures for dynamical systems modeling in adolescents: a feasibility and acceptability study. Translational behavioral medicine. 2016; 6(4):558±65. https://doi.org/10.1007/ s13142-016-0442-4 PMID: 27678501 33. Eysenbach G. The law of attrition. Journal of medical Internet research. 2005; 7(1):e11. https://doi.org/ 10.2196/jmir.7.1.e11 PMID: 15829473 34. Yesavage J, Brink T, Rose T. Geriatric depression scale (GDS). Handbook of psychiatric measures Washington DC: American Psychiatric Association. 2000:544±6. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 14 / 15 The Parkinson@Home study 35. Zhan A, Little MA, Harris DA, Abiola SO, Dorsey E, Saria S, et al. High Frequency Remote Monitoring of Parkinson's Disease via Smartphone: Platform Overview and Medication Response Detection. arXiv preprint arXiv:160100960. 2016. 36. Bot BM, Suver C, Neto EC, Kellen M, Klein A, Bare C, et al. The mPower study, Parkinson disease mobile data collected using ResearchKit. Scientific data. 2016; 3. 37. Etter J-F. Comparing the efficacy of two Internet-based, computer-tailored smoking cessation pro- grams: a randomized trial. Journal of Medical Internet Research. 2005; 7(1):e2. https://doi.org/10.2196/ jmir.7.1.e2 PMID: 15829474 38. Farvolden P, Denisoff E, Selby P, Bagby RM, Rudy L. Usage and longitudinal effectiveness of a Web- based self-help cognitive behavioral therapy program for panic disorder. Journal of medical Internet research. 2005; 7(1):e7. https://doi.org/10.2196/jmir.7.1.e7 PMID: 15829479 39. Hauser RA, Deckers F, Lehert P. Parkinson's disease home diary: further validation and implications for clinical trials. Movement Disorders. 2004; 19(12):1409±13. https://doi.org/10.1002/mds.20248 PMID: 40. Majority of the elderly in the Netherlands has a smartphone [Internet]. Houten, The Netherlands: Tele- compaper; 2015; 19 June 2015. https://www.telecompaper.com/pressrelease/majority-of-the-elderly- in-the-netherlands-has-a-smartphoneÐ1088067 41. Sullivan W. Trends on Tuesday: 186.3 million people own smartphones in the U.S. Digital Gov [Inter- net]. 2015 26-05-2015 https://www.digitalgov.gov/2015/05/26/trends-on-tuesday-186-3-million-people- own-smartphones-in-the-u-s/. 42. De Lau L, Giesbergen P, De Rijk M, Hofman A, Koudstaal P, Breteler M. Incidence of parkinsonism and Parkinson disease in a general population The Rotterdam Study. Neurology. 2004; 63(7):1240±4. PMID: 15477545 PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 15 / 15 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png PLoS ONE Unpaywall

Loading next page...
 
/lp/unpaywall/feasibility-of-large-scale-deployment-of-multiple-wearable-sensors-in-n4IyTlgOD0

References (56)

Publisher
Unpaywall
ISSN
1932-6203
DOI
10.1371/journal.pone.0189161
Publisher site
See Article on Publisher Site

Abstract

of large-scale deployment of multiple wearable sensors in Parkinson's disease. PLoS ONE 12(12): Wearable devices can capture objective day-to-day data about Parkinson's Disease (PD). e0189161. https://doi.org/10.1371/journal. This study aims to assess the feasibility of implementing wearable technology to collect data pone.0189161 from multiple sensors during the daily lives of PD patients. The Parkinson@home study is Editor: Oscar Arias-Carrion, Hospital General Dr. an observational, two-cohort (North America, NAM; The Netherlands, NL) study. To recruit Manuel Gea Gonzalez, MEXICO participants, different strategies were used between sites. Main enrolment criteria were Received: August 23, 2017 self-reported diagnosis of PD, possession of a smartphone and age18 years. Participants used the Fox Wearable Companion app on a smartwatch and smartphone for a minimum of Accepted: November 20, 2017 6 weeks (NAM) or 13 weeks (NL). Sensor-derived measures estimated information about Published: December 20, 2017 movement. Additionally, medication intake and symptoms were collected via self-reports in Copyright:© 2017 Silva de Lima et al. This is an the app. A total of 953 participants were included (NL: 304, NAM: 649). Enrolment rate was open access article distributed under the terms of 88% in the NL (n = 304) and 51% (n = 649) in NAM. Overall, 84% (n = 805) of participants the Creative Commons Attribution License, which permits unrestricted use, distribution, and contributed sensor data. Participants were compliant for 68% (16.3 hours/participant/day) of reproduction in any medium, provided the original the study period in NL and for 62% (14.8 hours/participant/day) in NAM. Daily accelerometer author and source are credited. data collection decreased 23% in the NL after 13 weeks, and 27% in NAM after 6 weeks. Data Availability Statement: The dataset used in Data contribution was not affected by demographics, clinical characteristics or attitude this article is owned by the Michael J Fox towards technology, but was by the platform usability score in the NL (χ (2) = 32.014, Foundation. In the original study consent forms, participants consented to share the full, de- p<0.001), and self-reported depression in NAM (χ (2) = 6.397, p = .04). The Parkinson@- identified study dataset only with qualified home study shows that it is feasible to collect objective data using multiple wearable sensors researchers. All qualified researchers in possession in PD during daily life in a large cohort. of a research question who would like to have access to the dataset used in this article, may do so through the Michael J. Fox Foundation Website (https://www.michaeljfox.org/page.html?id=192). A section of the website is devoted to sharing datasets, and the data used to support the PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 1 / 15 The Parkinson@Home study conclusions of this article will be available there. Introduction Interested and qualified researchers will be Parkinson's Disease (PD) is a common neurodegenerative disease in which patients experience required to register/apply and sign an Data Use both motor and non-motor symptoms[1]. Treatment is primarily based on the management Agreement that includes publication and IP policies. After registration, all enquiries for data of symptoms by increasing dopamine levels through pharmacological therapy or surgery[2, 3]. access will be judged by the Michael J Fox Additionally, non-pharmacological therapies, such as physiotherapy, occupational therapy or Foundation data sharing committee. speech therapy, are available to support patients[4]. Funding: This work was gratefully supported by Although good results in the management of motor symptoms have been achieved, particu- the Michael J Fox Foundation (https://www. larly in the early stages of the disease[5, 6], two major problems hamper long-term treatment. michaeljfox.org/), Intel Corporation (Tel Aviv, Israel First, current pharmacological therapy is successful for a limited period. In the long term, Ðhttps://www.intel.com) [grant number most patients develop unmanageable motor complications that can lead to worsening of qual- 10231.01], the Stichting Parkinson Fonds (https:// ity of life[7]. Second, evaluation of day-to-day variations in PD symptoms is difficult when www.parkinsonfonds.nl/), The Netherlands Organisation for Health Research and Development relying solely upon periodic consultations by clinicians[8]. Therefore, more detailed, objective (https://www.zonmw.nl/en/), Philips Research and reliable measures during daily living could potentially improve the management of PD. (https://www.philips.com/aw/research/home.html), Wearable sensors have been used to assess PD-related symptoms continuously and longitu- UCB pharma (http://www.ucb.com/), and the dinally during daily living[9±12]. Wearables may provide greater insight into a patient's Movement Disorders Society (https://www. disease status, allowing patients to self-manage their symptoms and monitor medication movementdisorders.org/MDS.htm). The author responses[13±18]. Furthermore, wearable sensor data may improve our scientific understand- Ana LõÂgia Silva de Lima is supported by Coordenac Ëão de Aperfeic Ëoamento de Pessoal de ing of disease progression by showing changes in motor and non-motor symptoms over time, NõÂvel SuperiorÐCAPES (http://www.capes.gov.br/) furthering the development of digital biomarkers for disease progression[19]. [grant number 0428-140]. The Michael J Fox While the potential value of wearable sensors for disease management and research are Foundation was involved in the data collection, increasingly becoming clear, various critical aspects of feasibility remain to be determined. interpretation of the data in the NAM cohort and Only a few studies have rigorously investigated the feasibility and acceptability of using a wear- manuscript revision. The funder Intel Corporation provided support in the form of salaries for authors able platform comprising a smartphone in combination to a smartwatch. Moreover, these EC and MA, participated in study design, data prior findings remained limited by the small sample sizes (biggest sample thus far: 40 PD collection and writing the report. The founders patients) [9, 13, 17, 18]. Therefore, we aimed to investigate the feasibility of using a wearable Philips Research and UCB pharma provided platform in a much larger sample of PD patients, with a focus on recruitment success, attrition support in the form of salaries for authors [Philips rates, user compliance and system usability. Research: HB; UCB pharma: KC, BB and DT] and participated in writing the report, but did not have any additional role in the study design, data Methods collection and analysis or decision to publish. The Between August/2015 and November/2016, a total of 953 PD patients from two cohorts specific roles of these authors are articulated in the (n = 304 in The Netherlands (NL) and n = 649 PD in North America (Unites States and Can- ªAuthor contributionsº section. All other sponsors did not have any role in study design; in the adaÐNAM) participated in the Parkinson@Home feasibility study. To investigate the feasibil- collection, analysis and interpretation of the data; in ity of the technology in different contexts, both cohorts used the same wearable platform, but the writing of the report; and in the decision to had distinct strategies for recruitment, retention and study period. These topics are described submit the paper for publication. separately (overview in Table 1). Competing interests: Ana LõÂgia Silva de Lima is supported by CoordenacËão de Aperfeic Ëoamento de Study design and population Pessoal de NõÂvel SuperiorÐCAPES [grant number 0428-140]. Tim Hahn is supported by a research The NL cohort. The population and study design applied in the NL are described in detail grant provided by the Stichting Parkinson Fonds. elsewhere.[20] In short, participants were recruited from support groups, internet communi- Luc JW Evers is supported by a research grant ties and through physiotherapists specialized in treating PD patients. Enrolment criteria provided by UCB and Philips Research. Nienke M were: (1)30 years of age, (2) possession of a smartphone using an Android OS version 4.2 de Vries is supported by a research grant from The and (3) self-reported diagnosis of PD. No exclusion criteria were applied beyond enrolment Netherlands Organisation for Health Research and criteria. Development. Eli Cohen and Michal Afek are employed by Intel. This affiliation does not alter our All enrolled participants received a single medical examination, based on the ªParkinson's adherence to PLOS ONE policies on sharing data Progression Markers Initiativeº (PPMI)[21]. This included the full MDS-UPDRS[22], the and materials. Lauren Bataille and Margaret Montreal Cognitive Assessment (MoCA)[23], and the Modified Schwab and England Activi- Daeschler are supported by the Michael J. Fox ties of Daily Living Scale.[24] The medical examination was performed by specially trained Foundation. Babak Boroojerdi, Kasper Claes and physiotherapists who are members of ParkinsonNet[25], a Dutch network of health PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 2 / 15 The Parkinson@Home study Dolors Terricabras are employed by UCB. This professionals specialised in PD management. At the end of the 13-weeks study period, all affiliation does not alter our adherence to PLOS enrolled participants evaluated the usability of the system through the System Usability Scale ONE policies on sharing data and materials. Max A (SUS)[26, 27], and were enquired about ability to use a smartphone (see APPENDIX). Finally, Little received research funding support from the participants had the option to continue using the platform or return the Pebble smartwatch. Michael J Fox Foundation and UCB. Heribert The NAM cohort. Study recruitment for the NAM cohort was entirely virtual through Baldus is employed by Philips Research. This affiliation does not alter our adherence to PLOS direct emails to subjects participating in the ªFox Insight online studyº, Facebook advertise- ONE policies on sharing data and materials. ments to targeted populations, and advertisements on Fox Trial Finder, a clinical trial match- Bastiaan R Bloem has served as an editorial board ing tool for people with PD.[28] Additional to the NL, the following enrolment criteria were member of Movement Disorders, currently serves applied: (1)18 years of age and (2) participation in the Fox Insight Online Study[29] as an editorial board member of Physiotherapy (Table 1). Canada, is Associate Editor for the Journal of In order to enrol, interested participants had to first register in the Fox Insight study (if they Parkinson's disease, received honoraria from serving on the scientific advisory board for Danone had not done so already). Through Fox Insight, each participant completed online surveys and Zambon, has received fees for speaking at about demographics, medical history, cognition, physical activity, symptoms and PD related conferences from AbbVie and Teva, and received medications and surgeries. Once enrolled in the Fox Insight, participants were eligible to regis- research support from the Netherlands ter for the NAM cohort of the Parkinson@Home study on a separate webpage. These users Organization for Scientific Research, the Michael J completed an online enrolment form which was reviewed by the study team to determine eligi- Fox Foundation, the Prinses Beatrix Foundation, the Stichting Parkinson Fonds, the National Parkinson bility. All study registrants received an email confirming their eligibility or non-eligibility. After Foundation and the Parkinson Vereniging. Marjan J finishing the 6-weeks study period, participants had the option to continue using the platform. Faber received grant support from the Michael J Fox Foundation, the Stichting Parkinson Fonds, Philips Research, The Netherlands Organisation for Wearable platform Health Research and Development and Health The Intel Pharma Analytics Platform used has been described in detail elsewhere.[20, 30] Holland. Briefly, it consists of the Fox Wearable Companion app, used on both a smartwatch and Table 1. Study design and procedure overview at the two cohorts. The Netherlands North America Recruitment strategies Through Internet communities ✓ ✓ Through support groups ✓ - Through physiotherapists ✓ - Enrolment criteria  30 years old ✓ - Dutch resident ✓ - Smartphone using Android OS version 4.2 or higher ✓ ✓ Self-reported PD ✓ ✓ 18 years old - ✓ Registered for Fox Insight study - ✓ English-speaking Canadian or United States resident - ✓ Exclusion criteria None ✓ ✓ Consent process Informative email ✓ ✓ Online digital consent form ✓ ✓ Study kit Pebble smartwatch ✓ ✓ Installation guide ✓ ✓ User manuals ✓ ✓ Clinical evaluations Assessment by physical therapist ✓ - Fox Insight online self-assessment surveys - ✓ Study duration Minimum of 6 weeks - ✓ Minimum of 13 weeks ✓ - Instruction for device usage Minimum of 5 hours a day - ✓ 24 hours, 7 days a week ✓ - Support model Call-center during working hours ✓ ✓ Technical support calls for non-data contributors ✓ - Support emails for non-contributors - ✓ Usability questionnaire - ✓ - https://doi.org/10.1371/journal.pone.0189161.t001 PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 3 / 15 The Parkinson@Home study Fig 1. (a) Fox Wearable Companion app main screen; (b) Fox Wearable Companion app activity graph; (c) Fox Wearable Companion app movement during sleep graph; (d) Fox Wearable Companion app symptom self-reports. ªReprinted from [Intel and Michael J Fox Foundation] under a CC BY license, with permission from [INTEL ], original copyright [2017]. https://doi.org/10.1371/journal.pone.0189161.g001 smartphone, and a cloud environment. In this study, a Pebble smartwatch, was used together with the patients' Android phones. 50 Hz accelerometer data were collected continuously from the smartwatch and streamed to the smartphone. Sensor analysis algorithms are applied to the aggregated (30 second interval) smartwatch accelerometer data in the app to estimate outcomes (i.e. levels of activity, tremor and move- ment during sleep). These estimated quantities are transmitted, via Wi-Fi or mobile data, to a cloud environment. They are also presented to the user by graphs and summary reports within the app. Additionally, users are able to set medication reminders, report actual medication intake and rate their symptoms (e.g. tremor, dyskinesia, rigidity, bradykinesia) within the mobile app (Fig 1). Both estimated outcomes and patients reported outcomes (PROs) are stored in the cloud environment. Study procedures at both cohorts Participants from both cohorts provided electronic consent and received a research kit con- taining a Pebble smartwatch, an installation guide and user manuals. Next, participants installed the Fox Wearable Companion App on their devices and were asked to wear the smartwatch and keep their smartphone with them as much as possible on either a 24/7 basis for 13-weeks study period (NL) or for a minimum of 5 hours a day, 7 days a week, for a 6-weeks study period (NAM). Additionally, participants reported their medication intake (i.e medication name and doses) and PD symptom severity using the app. A helpline was available during the study period for technical support. Support calls or emails were sent to participants from whom data were not collected for more than seven consecutive days. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 4 / 15 The Parkinson@Home study Outcome definitions and statistical analysis Feasibility assessment included recruitment, attrition, compliance and system usability. Recruitment success was analysed by (1) the total number of enrolled participants and (2) the number of eligible registrants that did not complete the informed consent. Compliance, simi- lar to previous studies[31, 32], was calculated as the median percentage of the study period where accelerometer data were collected. Attrition rate, based upon Eysenbach et al. [33], were measured by (1) decrease in the daily percentage of collected accelerometer data during each study period and (2) decrease in the number of participants contributing accelerometer data. Finally, system usability was measured by the median total score on the System Usability Scale. We investigated the relationship between self-reported demographics, clinical data, ability to use a smartphone (S1 File), System Usability score and the percentage of accelerometer data collected to identify factors that influence compliance levels. Participant demographic and clinical characteristics were grouped into categories either following previously described liter- ature (presence of depression[34]; presence of cognitive impairment[23]) or by convenience (age; educational level: a measure of the last completed level of education where low education was equal to high school or lower levels, middle education was equal to bachelor, and high education was equal to master or higher levels; Hoehn & Yahr stage and Modified Schwab and England scale). Because compliance was not normally distributed, the median and quartiles were used to divide participants into three compliance groups (low, middle and high). The first quartile was the cut-off for the low compliant group and third quartile for the high com- pliant group. Depending on the distribution of other variables in the analysis, either Chi- square, Fisher's Exact Test or Kruskal-Wallis were used to investigate significant differences between compliance groups considering demographics, clinical characteristics, ability to use a smartphone and System Usability score. Ethics standards This study was conducted in compliance with the Ethical Principles for Medical Research Involving Human Subjects, as defined in the Declaration of Helsinki. The study protocol and communication materials were approved by the local ethics committee (NL: CMO Arnhem- Nijmegen; NL53034.091.15; NAM: New England IRB: 15±046). Results Recruitment and sample characteristics In the NL cohort, 347 eligible PD patients were invited to participate. Among those invited, 43 refused to participate. The main refusal reasons were ªStudy protocol seems too burdensomeº (44%, n = 19), followed by ªPersonal circumstancesº (33%, n = 14). A total of 304 patients (enrolment rate = 88%) were enrolled. In the NAM cohort, from the 866 participants of the Fox Insight study who received a direct invitation to participate, 306 were enrolled (6% were ineligible). 344 additional participants were included from the remaining recruitment channels, with varied ineligibility rates. A total of 649 registrants (enrolment rate = 51%) were enrolled. In both cohorts, 953 participants were enrolled. From them, 805 were data contributors (participants that contributed at least one accelerometer data point during study period). Anal- ysis of the demographic characteristics of both cohorts showed that, in comparison to NA, the NL cohort presented more men (χ2 (1) = 9.5146, p<0.01); older (χ2 (2) = 16.435, p = 0.001) and higher educated (χ2 (2) = 25.270, p<0.001) PD included participants. The characteristics of all participants are presented in Table 2. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 5 / 15 The Parkinson@Home study Table 2. Demographic and disease related characteristics of participants. The Netherlands North America Total Data Non- p- Data Non- p- Data Non- contributors compliant value contributors compliant value contributors compliant (n = 291)* (n = 13)* (n = 514)* (n = 135)* (n = 805)* (n = 148)* Sex Men 168 (65%) - NA 260 (53%) 25 (51%) .763 454 (58%) 25 (51%) Age 50 29 (10%) 1 (8%) .523 101 (21%) 11 (22%) .676 130 (17%) 12 (19%) 51±69 207 (72%) 8 (61%) 310 (63%) 28 (57%) 517 (66%) 36 (58%) 70 53 (18%) 4 (31%) 81 (16%) 10 (20%) 134 (17%) 14 (23%) Education Level Low 51 (20%) NA NA 168 (35%) 21 (44%) .477 219 (30%) 21 (44%) Middle 103 (40%) NA NA 187 (39%) 16 (33%) 290 (39%) 16 (33%) High 101 (40%) NA NA 126 (26%) 11 (22%) 227 (31%) 11 (22%) Disease severity 0/1 73 (30%) 1 (100%) NA 235 (49%) 21 (47%) .580 308 (43%) 22 (48%) 2 127 (53%) 0 134 (28%) 10 (22%) 261 (36%) 10 (22%) 3 34 (14%) 0 98 (20%) 13 (29%) 132 (18%) 13 (29%) 4/5 6 (2%) 0 13 (3%) 1 (2%) 19 (3%) 1 (2%) Depression No 238 (97%) - NA 346 (70%) 26 (53%) .013 584 (79%) 26 (53%) Cognitive impairment No (>26) 124 (53%) 1 (100%) NA NA NA NA 124 (53%) 1 (100%) Independency level 70 36 (15%) 0 NA NA NA NA 36 (15%) 0 71±80 51 (22%) 0 NA NA 51 (22%) 0 81±90 109 (46%) 1 (100%) NA NA 109 (46%) 1 (100%) 91 41 (17%) 0 NA NA 41 (17%) 0 How easy is for you to Very easy 59(22%) 0 .211 NA NA NA 59(22%) 0 use a smartphone? Easy 117(44%) 3(50%) NA NA 117(44%) 3(50%) Neither easy 64(22%) 1 (17%) NA NA 64(22%) 1 (17%) nor difficult Difficult 20(7%) 2 (33%) NA NA 20(7%) 2 (33%) Very difficult 6(2%) NA NA 6(2%) MDS-UPDRS (Median) 52.5 (QR 35± 22 - NA NA 52.5 (QR 35± 22 69) 69) *Number of missing values differed across variables; only valid percentages are reported. SES: Socioeconomic status; Disease severity: assessed with Hoehn and Yarh stages at the NL cohort and estimated from self-reported at NAM cohort; MDS-UPDRS: Movement Disorders SocietyÐUnified Parkinson's Disease Rating Scale. -Depression was assessed with the Geriatric Depression Scale at NL and self-reported on NAM site; -Cognitive impairment was assessed using the Montreal Cognitive Assessment, -Independency level was measured by Modified Schwab and England Activities of Daily Living Scale. st rd NA = not assessed. QR: 1 and 3 quartiles. https://doi.org/10.1371/journal.pone.0189161.t002 Technical support to participants In both cohorts, the helpdesk consisted of two research assistants, available for 20 hours (NL) and 40 hours (NAM) per week. The actual workload was dependent on: (1) the number of participants simultaneously enrolled in the trial; and (2) the occurrence of bugs in the app or server downtime. The most frequent and time-consuming problems were: (1) Bluetooth dis- connection between the smartwatch and the smartphone and (2) questions regarding the med- ication report, especially in the first weeks of participation. Compliance Among both cohorts, 85% (n = 805 of 953 enrolled) of participants were data contributors. In the NL, 291 data-contributors collected data for a median of 1,478 hours each in the 13-weeks, PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 6 / 15 The Parkinson@Home study Fig 2. Distribution of compliance among all enrolled participants in the NL (n = 304-black) and NAM (n = 649-white) study cohorts. https://doi.org/10.1371/journal.pone.0189161.g002 st rd with quartile ranges (1 and 3 QR) of 888 to 1,827 hours. In NAM, 514 data contributors st rd collected a median of 621 hours (1 QR: 286 and 3 QR: 828 hours) each during the 6-weeks. st rd Compliance rates for each cohort were 68% (1 and 3 QR: 41%-83%) equal to 16.3 hours/ st rd participant/day in the NL and 62% (1 and 3 QR: 28%-82%) equal to 14.8 hours/participant/ day in NAM (Fig 2). Attrition In the NL, 13 participants (4% of all NL enrolled participants) did not contribute any data dur- ing the study period and were thus non-compliant. In NAM, this number was 135 (21% of all NAM enrolled participants). Additionally, 82 (27% of all enrolled) data-contributors in the NL became non-compliant during the study period. The primary known reasons (n = 47) were ªPersonal circumstancesº (38%, n = 18) and ªSystem too complex/System related issuesº (34%, n = 16). For the NAM cohort, although reasons were unknown, this number was 89 (17% of all enrolled). The attrition in the median percentage of sensor data collected daily varied between cohorts. In the NL, the attrition rate was 23% after 13-weeks' study period. In the NAM, attri- tion was 27% after 6-weeks' study period (Fig 3). Attrition in participation was tracked during and beyond the compulsory study period for each cohort. he number of participants decreased rapidly after the end of the study period in the NL cohort. A more gradual attrition in participation occurred in the NAM cohort (Fig 4). Ninety-six percent (n = 280) of data-contributors in the NL reported their medication through the app, while 78% (n = 404) did so in NAM. On average, data-contributors who used medication reports reported 351±217 medication intakes during the 13-week study period in PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 7 / 15 The Parkinson@Home study Fig 3. Attrition in compliance per day for NL (n = 291, black) and NAM participants (n = 514, gray) during the follow up period. https://doi.org/10.1371/journal.pone.0189161.g003 the NL and 127±113 over 6-week study period in NAM. Both cohorts showed a low and non- exponential attrition in medication report, similar to the attrition showed in compliance with the accelerometer data (data not shown). System usability In the NL cohort, 256 participants completed the System Usability Scale (response rate = 71.4%). st rd The median score was 62.5 (1 and 3 QT 47.5±72.5), which classifies the wearable platform in a category between ªOkº and ªGoodº (Fig 5). Factors related to compliance After grouping all NL data-contributors into compliance groups, analysis reveals no significant differences in the distribution of demographics, clinical characteristics and ability to use a smartphone between these groups. However, Kruskal-Wallis analysis demonstrates that the System Usability score reported is significantly different between the groups (χ (2) = 32.014, p<0.001). The mean rank score is 84.8 for the low compliant group, 130.8 for the middle com- pliant group and 160.0 for the high compliant group, which indicates that participants in the high compliant group provided a higher usability score to the system. For the NAM cohort, analysis shows that demographics and clinical characteristics between the three compliance groups was comparable, except for a trend regarding self-reported depression (χ (2) = 6.397, p = .04). This result indicates that a slightly higher number of self- reported depressed patients are in the low compliant group (Table 3). PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 8 / 15 The Parkinson@Home study End of follow-up End of follow-up Fig 4. Number of participants actively collecting sensor data at the NL (gray) and NAM (black) cohorts during and after the follow- up period (total initial n = 805). https://doi.org/10.1371/journal.pone.0189161.g004 Discussion This study assessed the feasibility of using a wearable platform for long-term data collection in a large sample of PD patients. We focused on: recruitment success, attrition rates, compliance Fig 5. SUS scoring of the Fox Wearable Companion platform (smartwatch with smartphone app) as rated by participants. https://doi.org/10.1371/journal.pone.0189161.g005 PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 9 / 15 The Parkinson@Home study Table 3. Distribution of data-contributors' characteristics and influence on compliance for the NL and NAM cohorts. The Netherlands North America Low Middle High p- Low Middle High p- compliance compliance compliance value compliance compliance compliance value (n = 73)* (n = 146)* (n = 72)* (n = 128)* (n = 256)* (n = 129)* 4 4 Sex Men 39 (71%) 84 (64%) 44 (65%) .63 64 (56%) 133 (54%) 62 (49%) .50 4 4 Age 50 7 (10%) 15 (10%) 7 (10%) .95 30 (26%) 49 (20%) 22 (17%) .41 51±69 51 (70%) 103 (71%) 53 (75%) 67 (57%) 161 (65%) 81 (64%) 70 15 (20%) 27 (19%) 11 (16%) 20 (17%) 36 (15%) 24 (19%) 4 4 Education Level Low 11 (20%) 23 (17%) 17 (25%) .68 46 (41%) 84 (35%) 38 (31%) .14 Middle 21 (38%) 45 (84%) 24 (35%) 43 (38%) 86 (35%) 57 (46%) High 23 (42%) 51 (39%) 27 (40%) 24 (21%) 73 (30%) 29 (23%) Disease severity 0+1 15 (42%) 37 (28%) 21 (30%) .55 50 (44%) 121 (50%) 64 (52%) .45 2 15 (42%) 72 (54%) 40 (57%) 30 (26%) 66 (27%) 38 (31%) 3 6 (16%) 20 (15%) 8 (11%) 30 (26%) 48 (19%) 19 (15%) 4+5 0 (0%) 5 (4%) 1 (1%) 4 (4%) 7 (3%) 2 (2%) 1 5 4 Depression No (<6) 36(100%) 135 (98%) 67 (94%) .28 74 (63.2%) 172 (70%) 99 (78%) .04 2 4 Cognitive impairment No (>26) 16 (47%) 74 (56%) 34 (50%) .58 - - - - Level of 70 5 (14%) 21 (16%) 10 (15%) .16 - - - - independency 71±80 13 (36%) 30 (23%) 8 (12%) - - - 81±90 13 (36%) 61 (46%) 35 (52%) - - - 91 5 (14%) 21 (16%) 15 (22%) - - - How easy is for you to Very easy 12 (19%) 33 (24%) 14 (21%) .06 - - - - use a smartphone? Easy 27 (44%) 60 (44%) 30 (44%) - - - Neither easy 14 (23%) 27 (20%) 23 (34%) - - - nor difficult Difficult 5 (8%) 14 (10%) 1 (2%) - - Very difficult 4 (6%) 2 (2%) (0%) - - - - MDS-UPDRS 49 55 48 .55 - - - - (median) SUS (median) 50 65 70 <0.001 - - - - *Missing values varied across variables. Red is significant at .05; green is significant at .001. SES: Socioeconomic status; Disease severity: assessed with Hoehn and Yarh stages at the NL cohort and estimated from self-reported at NAM cohort; GDS: Geriatric Depression Scale; MoCA: Montreal Cognitive Assessment, MDS-UPDRS: Movement Disorders SocietyÐUnified Parkinson's Disease Rating Scale, SUS: System Usability Scale. -Depression was assessed with the Geriatric Depression Scale at NL and self-reported on NAM; -Cognitive impairment was assessed using the Montreal Cognitive Assessment; -Independency level was measured by Modified Schwab and England Activities of Daily Living Scale; ±Pearson Chi-Square; ±Fisher's Exact Test; -Kruskal-Wallis https://doi.org/10.1371/journal.pone.0189161.t003 and system usability. Enrolment rate was 88% (n = 304) in the NL and 51% (n = 649) in NAM. Nearly 85% of all enrolled participants contributed sensor data during the study period. Median compliance rate was 68% (16.3 hours/participant/day) in the NL, and 62% (14.8 hours/participant/day) in NAM. The rate of accelerometer data collected each day declined 23% in the NL after 13-weeks of study period, and 27% in NAM after 6-weeks of study period. The distribution of demographics, clinical characteristics and ability to use a smartphone did not differ across compliance groups in the NL, but System Usability score did differ. For the PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 10 / 15 The Parkinson@Home study NAM, the distribution of demographics and clinical characteristics between the compliance groups was comparable, except for self-reported depression status. The high compliance in this study shows that it is feasible for people with PD to use this wearable platform in a real-world environment for many months. Although the feasibility of using consumer wearable sensors to monitor PD symptoms has been previous reported[9, 13, 17, 18, 31, 35], this is the first rigorous observational study to investigate the feasibility of a wear- able platform comprising a smartwatch combined with a smartphone in such a large patient group (the largest prior study included only 40 patients). Additionally, the small differences in study protocols across cohorts allowed us to observe the impact of varying usage instructions on compliance. Comparing the feasibility results obtained in this study to other studies, where either mobile apps were used in large cohorts[36] or e-health technologies were use[37, 38] d, we achieved a high compliance together with small and non-exponentially decreasing attrition rate, even though exponential decrease in compliance is the norm in these sort of studies[33]. This unusually high compliance rate may be attributed to the ªpassive” data collection. In this case, little or no interaction with the technology is required in order to collect sensor data. Participants using the Parkinson@Home wearable platform, other than reporting their medi- cation intake when reminded by the alarm (which was widely perceived as a service, instead of a burden), did not need to interact actively with the smartphone or smartwatch. In another similar smartphone-based study where ªactive”, ªtask-basedº monitoring was used (that is, where participants needed to perform certain specific tasks, at regular intervals prompted by the platform)[36], a more typical high and exponential attrition rate was observed. While it is difficult to draw firm conclusions from this comparison (because the two platforms are some- what different), we suspect that periodic and long interaction by users may increase attrition, leading to attrition rates seen in paper-based diaries[39]. The low and non-exponential attri- tion seen in the medication reports, a quick and less burdensome task, strengthened this conclusion. Thus, passive monitoring, where little to no interaction with the technology is required, may lead to better overall compliance rates. Despite the potential influence of age, gender and PD-related impairment (i.e. physical or cognitive) on compliance, our results showed that overall disease severity, MDS-UPDRS scores, independency level or cognitive impairment, did not influence compliance, which sug- gests that this platform could be used by most PD patients. The unique design of the Parkin- son@home study can partially explain this result. The presence of a personalized support centre, which was previously described as an effective strategy to improve retention of partici- pants[33], may have increased patients' confidence in using the system and have compensated for any disease-related difficulties. Moreover, the ªpro-activeº support model, with scheduled calls to participants who showed signs of low compliance, may have boosted compliance by providing a quick resolution of technical interruptions, and addressed any apathy towards participation caused by technology difficulties. This support is even more important because compliance is compromised in participants that reported low System Usability scores. Therefore, in order to achieve high compliance while using smartphone/smartwatch wearable platforms to measure PD related symptoms at home, it is beneficial to: (1) improve the platform's usability, (2) reduce the num- ber of technical issues, and (3) run a personalized support centre that can provide guidance to deal with possible technology related issues that participants may encounter. Limitations The Parkinson@Home study did have a few limitations. First, this is one of the first large-scale cohort studies using consumer wearable sensors in PD, with a long study period duration (i.e. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 11 / 15 The Parkinson@Home study up to 13 weeks). However, the study sample consisted only of PD patients that possessed a smartphone, thus introducing a possible selection bias, e.g. towards more highly educated sub- jects. Although smartphone penetration in the NL and NAM is high[40, 41], participants may not reflect the majority of PD patients living in the Netherlands, North America, or elsewhere in the world. Furthermore, when compared to the general PD population[42], participants were mainly young with a mild disease impairment and with some degree of cognitive impaired. Even though these variables showed no obvious influence on compliance, a more impaired population may need more personal support in order to maintain compliance. Future studies should aim for a more stratified population in order to further confirm the lack of influence across the full range of disease severity in the compliance with wearable sensors among PD patients. Second, the present results only apply to the use of two specific consumer grade devices (i.e. smartphone and smartwatch). Although consumer grade devices bring potential advantage over the use of dedicated medical devices, it is unknown whether our promising feasibility results would generalize to dedicated medical devices which are often more expensive and less user-friendly. Conclusion In conclusion, the Parkinson@home trial showed that it is feasible to deploy a technology plat- form consisting of consumer-grade wearable and mobile devices for long-term data collection in a large and geographically diverse PD population. Importantly, compliance was comparable for patients with a range of backgrounds, including men and women, different ages, and some variations in disease severity. These findings suggest that wearables may offer a promising approach to overcome the limitations in monitoring disease status and progression of mildly impaired PD patients in a real-life environment. The platform here used is a promising and practical approach to capturing large amounts of sensor data from many participants by pas- sive means, without much need for interaction with the technology. In the future, these prop- erties may position sensor technologies as effective tools for monitoring PD and the ªlived experienceº of PD patients. Supporting information S1 File. Scale created by the researchers to measure ability to use a smartphone. (TIF) Acknowledgments We thank the many people with Parkinson's disease who participated in these studies and the Stichting Parkinson Nederland. Without them this work would not have been possible. Author Contributions Conceptualization: Ana Lõgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Eli Cohen, Michal Afek, Lauren Bataille, Margaret Daeschler, Max A. Little, Bastiaan R. Bloem, Marjan J. Faber. Data curation: Ana Lõgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Nienke M. de Vries, Eli Cohen, Michal Afek, Lauren Bataille, Margaret Daeschler. Formal analysis: Ana Lõ Âgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Lauren Bataille, Marga- ret Daeschler, Marjan J. Faber. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 12 / 15 The Parkinson@Home study Funding acquisition: Bastiaan R. Bloem, Marjan J. Faber. Investigation: Ana Lõ Âgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Nienke M. de Vries, Lau- ren Bataille, Margaret Daeschler, Bastiaan R. Bloem, Marjan J. Faber. Methodology: Bastiaan R. Bloem, Marjan J. Faber. Project administration: Bastiaan R. Bloem, Marjan J. Faber. Resources: Lauren Bataille, Margaret Daeschler. Software: Eli Cohen, Michal Afek. Supervision: Bastiaan R. Bloem, Marjan J. Faber. Visualization: Ana Lõ Âgia Silva de Lima, Tim Hahn. Writing ± original draft: Ana Lõ Âgia Silva de Lima, Tim Hahn, Luc J. W. Evers. Writing ± review & editing: Ana Lõ Âgia Silva de Lima, Tim Hahn, Luc J. W. Evers, Nienke M. de Vries, Eli Cohen, Michal Afek, Lauren Bataille, Margaret Daeschler, Kasper Claes, Babak Boroojerdi, Dolors Terricabras, Max A. Little, Heribert Baldus, Bastiaan R. Bloem, Marjan J. Faber. References 1. Gelb DJ, Oliver E, Gilman S. Diagnostic criteria for Parkinson disease. Archives of neurology. 1999; 56 (1):33±9. PMID: 9923759 2. Goetz CG, Poewe W, Rascol O, Sampaio C. Evidence-based medical review update: Pharmacological and surgical treatments of Parkinson's disease: 2001 to 2004. Movement Disorders. 2005; 20(5):523± 39. https://doi.org/10.1002/mds.20464 PMID: 15818599 3. Mehanna R, Lai EC. Deep brain stimulation in Parkinson's disease. Translational neurodegeneration. 2013; 2(1):22. https://doi.org/10.1186/2047-9158-2-22 PMID: 24245947 4. Cutson TM, Laub KC, Schenkman M. Pharmacological and Nonpharmacological Interventions in the Treatment of Parkinson's Disease. Physical Therapy. 1995; 75(5):363±73. PMID: 7732080 5. Lewis S, Foltynie T, Blackwell A, Robbins T, Owen A, Barker R. Heterogeneity of Parkinson's disease in the early clinical stages using a data driven approach. Journal of Neurology, Neurosurgery & Psychia- try. 2005; 76(3):343±8. 6. Giugni JC, Okun MS. Treatment of advanced Parkinson's disease. Current opinion in neurology. 2014; 27(4):450±60. Epub 2014/07/01. https://doi.org/10.1097/WCO.0000000000000118 PMID: 24978634. 7. Keus SHJ, Munneke M, Graziano M, Jaana P. European Physiotherapy Guideline for Parkinson's dis- ease. KNGF/ParkinsonNet, the Netherlands. 2014. 8. Mera TO, Heldman DA, Espay AJ, Payne M, Giuffrida JP. Feasibility of home-based automated Parkin- son's disease motor assessment. Journal of neuroscience methods. 2012; 203(1):152±6. https://doi. org/10.1016/j.jneumeth.2011.09.019 PMID: 21978487 9. Arora S, Venkataraman V, Zhan A, Donohue S, Biglan K, Dorsey E, et al. Detecting and monitoring the symptoms of Parkinson's disease using smartphones: a pilot study. Parkinsonism & related disorders. 2015; 21(6):650±3. 10. Patel S, Chen BR, Mancinelli C, Paganoni S, Shih L, Welsh M, et al. Longitudinal monitoring of patients with Parkinson's disease via wearable sensor technology in the home setting. Conf Proc IEEE Eng Med Biol Soc. 2011; 2011:1552±5. Epub 2012/01/19. https://doi.org/10.1109/IEMBS.2011.6090452 PMID: 11. Maetzler W, Domingos J, Srulijes K, Ferreira JJ, Bloem BR. Quantitative wearable sensors for objective assessment of Parkinson's disease. Mov Disord. 2013; 28(12):1628±37. https://doi.org/10.1002/mds. 25628 PMID: 24030855. 12. Hobert MA, Maetzler W, Aminian K, Chiari L. Technical and clinical view on ambulatory assessment in Parkinson's disease. Acta Neurol Scand. 2014; 130(3):139±47. https://doi.org/10.1111/ane.12248 PMID: 24689772. 13. Tsanas A, Little MA, McSharry PE, Ramig L. Using the cellular mobile telephone network to remotely monitor parkinsons disease symptom severity. IEEE Transactions on Biomedical Engineering. 2012. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 13 / 15 The Parkinson@Home study 14. Patel S, Lorincz K, Hughes R, Huggins N, Growdon J, Standaert D, et al. Monitoring motor fluctuations in patients with Parkinson's disease using wearable sensors. IEEE transactions on information technol- ogy in biomedicine: a publication of the IEEE Engineering in Medicine and Biology Society. 2009; 13 (6):864±73. Epub 2009/10/23. 15. Griffiths RI, Kotschet K, Arfon S, Xu ZM, Johnson W, Drago J, et al. Automated assessment of bradyki- nesia and dyskinesia in Parkinson's disease. Journal of Parkinson's disease. 2012; 2(1):47±55. https:// doi.org/10.3233/JPD-2012-11071 PMID: 23939408 16. Tzallas AT, Tsipouras MG, Rigas G, Tsalikakis DG, Karvounis EC, Chondrogiorgi M, et al. PERFORM: A System for Monitoring, Assessment and Management of Patients with Parkinson's Disease. Sensors. 2014; 14(11):21329±57. https://doi.org/10.3390/s141121329 PMID: 25393786 17. Lakshminarayana R, Wang D, Burn D, Chaudhuri KR, Cummins G, Galtrey C, et al. Smartphone-and internet-assisted self-management and adherence tools to manage Parkinson's disease (SMART-PD): study protocol for a randomised controlled trial (v7; 15 August 2014). Trials. 2014; 15(1):374. 18. Sharma V, Mankodiya K, De La Torre F, Zhang A, Ryan N, Ton TG, et al., editors. SPARK: personal- ized parkinson disease interventions through synergy between a smartphone and a smartwatch. Inter- national Conference of Design, User Experience, and Usability; 2014: Springer. 19. Espay AJ, Bonato P, Nahab FB, Maetzler W, Dean JM, Klucken J, et al. Technology in Parkinson's dis- ease: Challenges and opportunities. Movement Disorders. 2016; 31(9):1272±82. https://doi.org/10. 1002/mds.26642 PMID: 27125836 20. de Lima ALS, Hahn T, de Vries NM, Cohen E, Bataille L, Little MA, et al. Large-Scale Wearable Sensor Deployment in Parkinson's Patients: The Parkinson@ Home Study Protocol. JMIR Research Protocols. 2016; 5(3). 21. Marek K, Jennings D, Lasch S, Siderowf A, Tanner C, Simuni T, et al. The parkinson progression marker initiative (PPMI). Progress in neurobiology. 2011; 95(4):629±35. https://doi.org/10.1016/j. pneurobio.2011.09.005 PMID: 21930184 22. Goetz CG, Tilley BC, Shaftman SR, Stebbins GT, Fahn S, Martinez-Martin P, et al. Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): Scale pre- sentation and clinimetric testing results. Movement disorders. 2008; 23(15):2129±70. https://doi.org/10. 1002/mds.22340 PMID: 19025984 23. Hoops S, Nazem S, Siderowf A, Duda J, Xie S, Stern M, et al. Validity of the MoCA and MMSE in the detection of MCI and dementia in Parkinson disease. Neurology. 2009; 73(21):1738±45. https://doi.org/ 10.1212/WNL.0b013e3181c34b47 PMID: 19933974 24. Schwab R, England A. Projection Technique for Evaluating Surgery in Parkinson's Disease: Third Sym- posium on Parkinson's Disease. Edinburgh: Gillingham and Donaldson. 1969. 25. Bloem BR, Munneke M. Revolutionising management of chronic disease: the ParkinsonNet approach. BMJ. 2014; 348:g1838. https://doi.org/10.1136/bmj.g1838 PMID: 24647365 26. Bangor A, Kortum P, Miller J. Determining what individual SUS scores mean: Adding an adjective rating scale. Journal of usability studies. 2009; 4(3):114±23. 27. Brooke J. SUS-A quick and dirty usability scale. Usability evaluation in industry. 1996; 189(194):4±7. 28. Meunier CC, Chowdhury S, Sherer T, Brooks DW. Innovative Web-based Matching Service, Fox Trial Finder, as a mechanism to improve clinical trial recruitment. Movement Disorders. 2012; 27(4):e2. 29. Heusinkveld L, Hacker M, Turchan M, Bollig M, Tamargo C, Fisher W, et al. Patient Perspectives on Deep Brain Stimulation Clinical Research in Early Stage Parkinson's Disease. Journal of Parkinson's disease. 2017; 7(1):89±94. https://doi.org/10.3233/JPD-161031 PMID: 27911344 30. Cohen S, Bataille LR, Martig AK. Enabling breakthroughs in Parkinson's disease with wearable technol- ogies and big data analytics. mHealth. 2016; 2:20. Epub 2017/03/16. https://doi.org/10.21037/mhealth. 2016.04.02 PMID: 28293596. 31. Ferreira JJ, Godinho C, Santos AT, Domingos J, Abreu D, Lobo R, et al. Quantitative home-based assessment of Parkinson's symptoms: The SENSE-PARK feasibility and usability study. BMC neurol- ogy. 2015; 15(1):89. 32. Brannon EE, Cushing CC, Crick CJ, Mitchell TB. The promise of wearable sensors and ecological momentary assessment measures for dynamical systems modeling in adolescents: a feasibility and acceptability study. Translational behavioral medicine. 2016; 6(4):558±65. https://doi.org/10.1007/ s13142-016-0442-4 PMID: 27678501 33. Eysenbach G. The law of attrition. Journal of medical Internet research. 2005; 7(1):e11. https://doi.org/ 10.2196/jmir.7.1.e11 PMID: 15829473 34. Yesavage J, Brink T, Rose T. Geriatric depression scale (GDS). Handbook of psychiatric measures Washington DC: American Psychiatric Association. 2000:544±6. PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 14 / 15 The Parkinson@Home study 35. Zhan A, Little MA, Harris DA, Abiola SO, Dorsey E, Saria S, et al. High Frequency Remote Monitoring of Parkinson's Disease via Smartphone: Platform Overview and Medication Response Detection. arXiv preprint arXiv:160100960. 2016. 36. Bot BM, Suver C, Neto EC, Kellen M, Klein A, Bare C, et al. The mPower study, Parkinson disease mobile data collected using ResearchKit. Scientific data. 2016; 3. 37. Etter J-F. Comparing the efficacy of two Internet-based, computer-tailored smoking cessation pro- grams: a randomized trial. Journal of Medical Internet Research. 2005; 7(1):e2. https://doi.org/10.2196/ jmir.7.1.e2 PMID: 15829474 38. Farvolden P, Denisoff E, Selby P, Bagby RM, Rudy L. Usage and longitudinal effectiveness of a Web- based self-help cognitive behavioral therapy program for panic disorder. Journal of medical Internet research. 2005; 7(1):e7. https://doi.org/10.2196/jmir.7.1.e7 PMID: 15829479 39. Hauser RA, Deckers F, Lehert P. Parkinson's disease home diary: further validation and implications for clinical trials. Movement Disorders. 2004; 19(12):1409±13. https://doi.org/10.1002/mds.20248 PMID: 40. Majority of the elderly in the Netherlands has a smartphone [Internet]. Houten, The Netherlands: Tele- compaper; 2015; 19 June 2015. https://www.telecompaper.com/pressrelease/majority-of-the-elderly- in-the-netherlands-has-a-smartphoneÐ1088067 41. Sullivan W. Trends on Tuesday: 186.3 million people own smartphones in the U.S. Digital Gov [Inter- net]. 2015 26-05-2015 https://www.digitalgov.gov/2015/05/26/trends-on-tuesday-186-3-million-people- own-smartphones-in-the-u-s/. 42. De Lau L, Giesbergen P, De Rijk M, Hofman A, Koudstaal P, Breteler M. Incidence of parkinsonism and Parkinson disease in a general population The Rotterdam Study. Neurology. 2004; 63(7):1240±4. PMID: 15477545 PLOS ONE | https://doi.org/10.1371/journal.pone.0189161 December 20, 2017 15 / 15

Journal

PLoS ONEUnpaywall

Published: Dec 20, 2017

There are no references for this article.