scispace - formally typeset

Book ChapterDOI

From wellness to medical diagnostic apps: the Parkinson's Disease case

01 Jan 2017-pp 384-389

TL;DR: The design and development of the CloudUPDRS app and supporting system developed as a Class I medical device to assess the severity of motor symptoms for Parkinson’s Disease is presented.

AbstractThis paper presents the design and development of the CloudUPDRS app and supporting system developed as a Class I medical device to assess the severity of motor symptoms for Parkinson’s Disease. We report on lessons learnt towards meeting fidelity and regulatory requirements; effective procedures employed to structure user context and ensure data quality; a robust service provision architecture; a dependable analytics toolkit; and provisions to meet mobility and social needs of people with Parkinson’s.

Topics: Analytics (52%)

Summary (2 min read)

1 Introduction

  • It is well understood that modern smartphones present unique opportunities for mobile healthcare.
  • Yet, the vast majority of these apps do not conform to the safety, quality, performance and regulatory requirements set for medical devices and as such they can only be employed either to encourage a healthy lifestyle or for research purposes correspondingly, but are not tools for medical diagnosis.
  • In contrast to this situation, this paper presents the design and development of the CloudUPDRS app and its associated information management and analytics platform, which meets the standards set for medical devices.
  • In particular, the authors describe how CloudUPDRS achieves the accurate, precise, and repeatable assessment of motor symptoms for people with Parkinson’s (PwP), which clinicians can use with confidence.
  • In this paper the authors present the key contributions of this work towards achieving this goal.

2 Background and Rationale

  • PD symptoms are typically caused by the loss of neurones that produce dopamine, a key chemical messenger in the brain, decreased levels of which lead to abnormal brain activity (cf. [1] for more details).
  • Since symptoms vary greatly independent of treatment and PD progresses at different rates in different individuals, it requires regular clinical monitoring and medication adjustment.
  • Monitoring and adjustment however require hospital visits and assessment under the standard Movement Disorders Society’s Unified Parkinson’s Disease Rating Scale (mds-updrs) [2].
  • Due to these constraints, such reviews are relatively infrequent, carried out typically only a few times per year.
  • By adopting this approach, the authors also intend to capture in-depth medical intelligence supporting the discovery of longitudinal trends, promoting deeper understanding of the patterns of normal daily symptom variations, and predicting the onset of dyskinesias thus facilitating high-precision personalised targeting of treatment.

3 The CloudUPDRS app

  • Specifically, the authors designed, developed and validated in a field study a prototype app on Android implementing Part III of the mds-updrs.
  • Participants were also tested in the same areas of motor performance using the standard lab procedure outlined in mds-updrs and using bespoke biomedical data acquisition equipment to obtain a baseline for comparing the performance of the app.
  • The app implements a comprehensive workflow partially depicted in Figure 1, which provides audio, video and textual guidance on how to conduct the actions required by the tests and automatically adapts to match the specifications of its host device.
  • Overall, the CloudUPDRS system consists of the following elements: 1. PD patient smartphone apps for Android and iOS that carry out motor performance measurements and wellness self-assessment; conduct session management; securely transfer captured data to the CloudUPDRS service; and, present an interface providing guidance and feedback.
  • Data-mining toolkit for medical intelligence incorporating quantitative and semi-structured data, and longitudinal analyses, clustering and classification; and a clinical user interface incorporating visualisation.

4 The CloudUPDRS Service Platform

  • The CloudUPDRS service platform enables the secure capture, management and analysis of data collected by the app and provides effective communication of insights generated to clinicians enabling them to explore alternative treatment scenarios.
  • The microservices architectural style is set in contrast to traditional monolithic web applications and aims to maximise opportunities for vertical decomposition and scaling-out, which are critical for high performance and service resilience in data intensive situations.
  • In CloudUPDRS, microservices are loosely coupled and employ lightweight communication and coordination mechanisms such as the Consumer-Driven Contract pattern and implemented on Apache Thrift (cf. https://thrift.apache. org/) selected due to its highly efficient and compact protocol structure.
  • System componentization follows the design displayed in Figure 2, enforced via versioning of published RESTful interfaces.
  • While the data collection and signal processing APIs are implemented using python and django REST within an nginx/gunicorn container, semi-structured longitudinal analytics are implemented as Ruby bundles.

5 Lessons Learnt and Conclusions

  • The pervasive computing community has invested significant effort in techniques for modelling and adapting to user context, which is critical for the interpretation of sensor data streams.
  • Yet, when context modelling is not possible or incurs prohibitively high costs, an effective alternative is to bound context by imposing structure and thus predictability to user actions during sensing, an approach that was successfully applied with the CloudUPDRS app.
  • While there are clearly situations when the development of new algorithms and techniques is required, in other cases there seems to be good reason to opt for a more traditional approach.
  • In CloudUPDRS rather than optimise individual stages the authors engineer an end-to-end quality assurance strategy that they find to be more effective.
  • It incorporates features of the user experience, which permit the user to initiate the repeat of tests when an external event has disrupted the session, to increasing the duration of individual tests so as to enable oversampling and cross-validation, to employing heuristics that allow us to quickly identify data quality problems in the captured signal.

Did you find this useful? Give us your feedback

...read more

Content maybe subject to copyright    Report

BIROn - Birkbeck Institutional Research Online
Kueppers, Stefan and Daskalopoulos, I. and Jha, A. and Fragopanagos,
N.F. and Kassavetis, P. and Nomikou, E. and Saifee, T. and Rothwell, J.C.
and Bhatia, K. and Luchini, M.U. and Iannone, M. and Moussouri, T. and
Roussos, George (2016) From wellness to medical diagnostic apps: the
Parkinson’s Disease case. In: Giokas, K. and Bokor, L. and Hopfgartner,
F. (eds.) eHealth 360: International Summit on eHealth, Budapest,
Hungary, June 14-16, 2016, Revised Selected Papers. Lecture Notes of the
Institute for Computer Sciences, Social Informatics and Telecommunications
Engineering 181. New York, U.S.: Springer, pp. 384-389. ISBN
9783319496542.
Downloaded from:
Usage Guidelines:
Please refer to usage guidelines at or alternatively
contact lib-eprints@bbk.ac.uk.

From Wellness to Medical Diagnostic apps:
The Parkinson’s Disease Case
S. Kueppers
21
, I. Daskalopoulos
1
, A. Jha
3
, N.F. Fragopanagos
4
, P.
Kassavetis
36
, E. Nomikou
5
,, T. Saifee
3
, J.C. Rothwell
3
, K. Bhatia
3
, M.U.
Luchini
2
, M. Iannone
4
, T. Moussouri
53
, and G. Roussos
1
No Institute Given
Abstract. This paper presents the design and development of the
CloudUPDRS app and supporting system developed as a Class I medical
device to assess the severity of motor symptoms for Parkinson’s Disease.
We report on lessons learnt towards meeting fidelity and regulatory re-
quirements; effective procedures employed to structure user context and
ensure data quality; a robust service provision architecture; a depend-
able analytics toolkit; and provisions to meet mobility and social needs
of people with Parkinson’s.
1 Introduction
It is well understood that modern smartphones present unique opportunities for
mobile healthcare. Indeed, there are numerous wellness and self-tracking apps
readily available in all major mobile phone platform markets and many more
have been developed to conduct research in various aspects of mobile telecare.
Yet, the vast majority of these apps do not conform to the safety, quality, perfor-
mance and regulatory requirements set for medical devices and as such they can
only be employed either to encourage a healthy lifestyle or for research purposes
correspondingly, but are not tools for medical diagnosis. This fact is often ex-
plicitly reflected in their terms and conditions of use for example, quoting from
a popular Parkinson’s Disease app, the developers state that “we cannot, and
thus we do not, guarantee or promise that you will personally receive any direct
benefits.” In contrast to this situation, this paper presents the design and de-
velopment of the CloudUPDRS app and its associated information management
and analytics platform, which meets the standards set for medical devices. In
particular, we describe how CloudUPDRS achieves the accurate, precise, and
repeatable assessment of motor symptoms for people with Parkinson’s (PwP),
which clinicians can use with confidence. The app is currently undergoing exam-
ination by the Medicines and Healthcare products Regulatory Agency (MHRA)
in the UK towards its full registration as a medical device.
The successful development and operational deployment of the CloudUP-
DRS app and its supporting service at the level required to achieve conformal
performance to medical device regulations, thus establishing it as a valuable di-
agnostic tool for clinicians, demanded that we address several key problems. In

2 Authors Suppressed Due to Excessive Length
Fig. 1. Views of the user interface of the CloudUPDRS app showing session manage-
ment, tremor recording and finger tapping activities..
this paper we present the key contributions of this work towards achieving this
goal. Specifically, we describe:
How to effectively combine a guided data collection procedure imposed by the
app to provide structured user context, with a fully automated signal process-
ing pipeline thus making possible the unsupervised but consistent interpre-
tation of sensor data captured during the performance of motor assessment
activities.
The development of a data analytics toolkit for the assessment of tremor,
bradykenesia and gait measurements following the MDS Unified Parkinson’s
Disease Rating Scale, the standard clinical tool for the diagnosis of PD.
The development of an information management, data mining and dashboard
service developed following the concept of microservices and the lambda archi-
tecture, incorporating stream and batch processing pathways to ensure scale
out performance and responsiveness.
2 Background and Rationale
PD symptoms are typically caused by the loss of neurones that produce dopamine,
a key chemical messenger in the brain, decreased levels of which lead to abnor-
mal brain activity (cf. [1] for more details). Care for patients with PD involves
the management of both motor and non-motor symptoms as well as palliative
care.
Since symptoms vary greatly independent of treatment and PD progresses at
different rates in different individuals, it requires regular clinical monitoring and
medication adjustment. Monitoring and adjustment however require hospital
visits and assessment under the standard Movement Disorders Society’s Unified
Parkinson’s Disease Rating Scale (mds-updrs) [2]. Due to these constraints,
such reviews are relatively infrequent, carried out typically only a few times per
year. This in turn limits opportunities to precisely quantify PD progression and
the effectiveness of patient stratification [8]: the restricted availability of data
concerning individual variability and actual symptom trends limit opportunities
to adapt care to the needs of a particular individual at a specific time.

From Wellness to Medical Diagnostic apps: The Parkinson’s Disease Case 3
Indeed, it is possible to employ certain aspects of movement that are dis-
rupted in Parkinson’s as surrogate biomarkers of dopamine levels and in fact
this is precisely the purpose of Part III of the mds-updrs. Further pursuing this
insight, in [3] we investigated the possibility to precisely quantify and implement
the mds-updrs methodology as a smartphone app to enable the assessment of
motor performance through tremor, gait and bradykinesia measurements ob-
tained from standard sensors embedded in smartphones within a clinical set-
ting. By adopting this approach, we also intend to capture in-depth medical
intelligence supporting the discovery of longitudinal trends, promoting deeper
understanding of the patterns of normal daily symptom variations, and predict-
ing the onset of dyskinesias thus facilitating high-precision personalised targeting
of treatment.
3 The CloudUPDRS app
As discussed in the previous Section, in [3] we demonstrate the feasibility of
using smartphones as a means to assess commonly occurring motor symptoms
of PD in a clinical setting. Specifically, we designed, developed and validated in a
field study a prototype app on Android implementing Part III of the mds-updrs.
Using the accelerometer and touch screen sensors commonly available in modern
smartphones, we are able to carry out hand and leg tremor measurements, as
well as gait and bradykinesia assessments using finger tapping tasks to replicate
the majority of these tests. In [3] tests were administered by an experienced
clinician in the lab using an HTC Desire device and the collected sensor data
were extracted and processed using standard biomedical data analysis software.
Participants were also tested in the same areas of motor performance using the
standard lab procedure outlined in mds-updrs and using bespoke biomedical
data acquisition equipment to obtain a baseline for comparing the performance
of the app.
In CloudUPDRS we employ the data collection and analysis techniques de-
scribed in [3] to develop an app with extended functionality that enables its
independent but dependable use by PwP and their carers at home and in their
communities. The app implements a comprehensive workflow partially depicted
in Figure 1, which provides audio, video and textual guidance on how to con-
duct the actions required by the tests and automatically adapts to match the
specifications of its host device. The app is also provisioned with a delay toler-
ant background service to manage session data that ensures that information is
safely submitted for further processing to a supporting online service also devel-
oped specifically to provide this function and described in more detail in Section
4 below.
Overall, the CloudUPDRS system consists of the following elements:
1. PD patient smartphone apps for Android and iOS that carry out motor per-
formance measurements and wellness self-assessment; conduct session man-
agement; securely transfer captured data to the CloudUPDRS service; and,
present an interface providing guidance and feedback.

4 Authors Suppressed Due to Excessive Length
2. Cloud-based scalable dat collection engine that safely and securely collects
data from patients’ smartphones; ensures secure data management; and ap-
plies the mds-updrs processing pipeline.
3. Data-mining toolkit for medical intelligence incorporating quantitative and
semi-structured data, and longitudinal analyses, clustering and classification;
and a clinical user interface incorporating visualisation.
4 The CloudUPDRS Service Platform
The CloudUPDRS service platform enables the secure capture, management and
analysis of data collected by the app and provides effective communication of
insights generated to clinicians enabling them to explore alternative treatment
scenarios. To cater for the diverse needs of the PwP population in the UK,
the platform has been engineered to facilitate scalable performance by adopting
the microservices architecture [9]. The microservices architectural style is set
in contrast to traditional monolithic web applications and aims to maximise
opportunities for vertical decomposition and scaling-out, which are critical for
high performance and service resilience in data intensive situations.
In CloudUPDRS, microservices are loosely coupled and employ lightweight
communication and coordination mechanisms such as the Consumer-Driven Con-
tract pattern and implemented on Apache Thrift (cf. https://thrift.apache.
org/) selected due to its highly efficient and compact protocol structure. System
componentization follows the design displayed in Figure 2, enforced via version-
ing of published RESTful interfaces. CloudUPDRS microservices are deployed
as docker containers (cf. https://www.docker.com/) although internal imple-
mentation details vary to match the specific preferences and expertise of project
partners responsible for their implementation and their suitability for the task in
hand. For example, while the data collection and signal processing APIs are im-
plemented using python and django REST within an nginx/gunicorn container,
semi-structured longitudinal analytics are implemented as Ruby bundles.
Finally, the service platform has been designed with the expectation that
in order to meet performance metrics for its interactive features at full opera-
tion scale it will require the on the fly integration of archived information from
its longitudinal datastore with real-time streams captured for example during
concurrent patient consultations. To facilitate this modus operandi, we have
structured workflows implemented through microservices following the lambda
architecture [7], which provides an intuitive model for the fusion of both types
of data on the fly.
5 Lessons Learnt and Conclusions
Bounded Context. The pervasive computing community has invested significant
effort in techniques for modelling and adapting to user context, which is crit-
ical for the interpretation of sensor data streams. This role for context was

References
More filters

Journal ArticleDOI
Christopher G. Goetz1, Barbara C. Tilley2, Stephanie R. Shaftman2, Glenn T. Stebbins1, Stanley Fahn3, Pablo Martinez-Martin, Werner Poewe4, Cristina Sampaio5, Matthew B. Stern6, Richard Dodel7, Bruno Dubois8, Robert G. Holloway9, Joseph Jankovic10, Jaime Kulisevsky11, Anthony E. Lang12, Andrew J. Lees13, Sue Leurgans1, Peter A. LeWitt14, David L. Nyenhuis15, C. Warren Olanow16, Olivier Rascol17, Anette Schrag13, Jeanne A. Teresi3, Jacobus J. van Hilten18, Nancy R. LaPelle19, Pinky Agarwal, Saima Athar, Yvette Bordelan, Helen Bronte-Stewart, Richard Camicioli, Kelvin L. Chou, Wendy Cole, Arif Dalvi, Holly Delgado, Alan Diamond, Jeremy P.R. Dick, John E. Duda, Rodger J. Elble, Carol Evans, V. G. H. Evidente, Hubert H. Fernandez, Susan H. Fox, Joseph H. Friedman, Robin D. Fross, David A. Gallagher, Deborah A. Hall, Neal Hermanowicz, Vanessa K. Hinson, Stacy Horn, Howard I. Hurtig, Un Jung Kang, Galit Kleiner-Fisman, Olga Klepitskaya, Katie Kompoliti, Eugene C. Lai, Maureen L. Leehey, Iracema Leroi, Kelly E. Lyons, Terry McClain, Steven W. Metzer, Janis M. Miyasaki, John C. Morgan, Martha Nance, Joanne Nemeth, Rajesh Pahwa, Sotirios A. Parashos, Jay S. Schneider, Kapil D. Sethi, Lisa M. Shulman, Andrew Siderowf, Monty Silverdale, Tanya Simuni, Mark Stacy, Robert Malcolm Stewart, Kelly L. Sullivan, David M. Swope, Pettaruse M. Wadia, Richard Walker, Ruth H. Walker, William J. Weiner, Jill Wiener, Jayne R. Wilkinson, Joanna M. Wojcieszek, Summer C. Wolfrath, Frederick Wooten, Allen Wu, Theresa A. Zesiewicz, Richard M. Zweig 
TL;DR: The combined clinimetric results of this study support the validity of the MDS‐UPDRS for rating PD.
Abstract: We present a clinimetric assessment of the Movement Disorder Society (MDS)-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS). The MDS-UDPRS Task Force revised and expanded the UPDRS using recommendations from a published critique. The MDS-UPDRS has four parts, namely, I: Non-motor Experiences of Daily Living; II: Motor Experiences of Daily Living; III: Motor Examination; IV: Motor Complications. Twenty questions are completed by the patient/caregiver. Item-specific instructions and an appendix of complementary additional scales are provided. Movement disorder specialists and study coordinators administered the UPDRS (55 items) and MDS-UPDRS (65 items) to 877 English speaking (78% non-Latino Caucasian) patients with Parkinson's disease from 39 sites. We compared the two scales using correlative techniques and factor analysis. The MDS-UPDRS showed high internal consistency (Cronbach's alpha = 0.79-0.93 across parts) and correlated with the original UPDRS (rho = 0.96). MDS-UPDRS across-part correlations ranged from 0.22 to 0.66. Reliable factor structures for each part were obtained (comparative fit index > 0.90 for each part), which support the use of sum scores for each part in preference to a total score of all parts. The combined clinimetric results of this study support the validity of the MDS-UPDRS for rating PD.

3,385 citations


"From wellness to medical diagnostic..." refers methods in this paper

  • ...Monitoring and adjustment however require hospital visits and assessment under the standard Movement Disorders Society’s Unified Parkinson’s Disease Rating Scale (mds-updrs) [2]....

    [...]


Book
10 May 2015
TL;DR: Big Data describes a scalable, easy to understand approach to big data systems that can be built and run by a small team that takes advantage of clustered hardware along with new tools designed specifically to capture and analyze web-scale data.
Abstract: Services like social networks, web analytics, and intelligent e-commerce often need to manage data at a scale too big for a traditional database As scale and demand increase, so does Complexity Fortunately, scalability and simplicity are not mutually exclusiverather than using some trendy technology, a different approach is needed Big data systems use many machines working in parallel to store and process data, which introduces fundamental challenges unfamiliar to most developers Big Data shows how to build these systems using an architecture that takes advantage of clustered hardware along with new tools designed specifically to capture and analyze web-scale data It describes a scalable, easy to understand approach to big data systems that can be built and run by a small team Following a realistic example, this book guides readers through the theory of big data systems, how to use them in practice, and how to deploy and operate them once they're built Purchase of the print book comes with an offer of a free PDF, ePub, and Kindle eBook from Manning Also available is all code from the book

882 citations


"From wellness to medical diagnostic..." refers methods in this paper

  • ...To facilitate this modus operandi, we have structured workflows implemented through microservices following the lambda architecture [7], which provides an intuitive model for the fusion of both types of data on the fly....

    [...]


Journal ArticleDOI
TL;DR: Maintaining good motor function and quality of life remain the primary goals of therapy and the principle that treatment must be tailored to the individual patient’s needs is paramount.
Abstract: The predominant motor features of Parkinson's disease (PD) are caused by degeneration of dopaminergic neurones and can be reversed in part or whole by dopamine replacement or augmentation strategies. Physicians have most experience with the use of levodopa, which remains the most potent oral dopaminergic treatment for PD. There are reservations about the long-term use of levodopa, most particularly in the context of its propensity to induce motor fluctuations and dyskinesias. Strategies exist to delay or diminish these complications, but the physician must lay the basis for these in the selection of drugs for early treatment and the sequence of drugs introduced subsequently. Levodopa efficacy and duration of effect may be enhanced by combination with a catechol-O-methyl transferase inhibitor. Maintaining good motor function and quality of life remain the primary goals of therapy and the principle that treatment must be tailored to the individual patient's needs is paramount.

162 citations


Proceedings ArticleDOI
07 Mar 2011
TL;DR: This paper presents a novel method to estimate stride length through the application of the wavelet transform to the signal obtained from a wireless accelerometer on the waist, and introduces a novel metric to determine the level of theWavelet transform detail coefficients from which the step frequency can be directly extracted.
Abstract: Gait analysis using wireless accelerometers deployed as body area networks can provide valuable information for multiple health-related applications. Within this field, stride length estimation represents a difficult task. In this paper we present a novel method to estimate stride length through the application of the wavelet transform to the signal obtained from a wireless accelerometer on the waist. We also introduce a novel metric to determine the level of the wavelet transform detail coefficients from which the step frequency can be directly extracted. Additionally, we show the correlation between the energy of the wavelet transform approximation coefficients and the speed of the gait.

49 citations


Journal ArticleDOI
TL;DR: The first report on the development and testing of stand‐alone software for mobile devices that could be used to assess both tremor and bradykinesia of PD patients is provided.
Abstract: BACKGROUND: The natural fluctuation of motor symptoms of Parkinson's disease (PD) makes judgement of any change challenging and the use of clinical scales such as the International Parkinson and Movement Disorder Society (MDS)-UPDRS imperative. Recently developed commodity mobile communication devices, such as smartphones, could possibly be used to assess motor symptoms in PD patients in a convenient way with low cost. We provide the first report on the development and testing of stand-alone software for mobile devices that could be used to assess both tremor and bradykinesia of PD patients. METHODS: We assessed motor symptoms with a custom-made smartphone application in 14 patients and compared the results with their MDS-UPDRS scores. RESULTS: We found significant correlation between five subscores of MDS-UPDRS (rest tremor, postural tremor, pronation-supination, leg agility, and finger tapping) and eight parameters of the data collected with the smartphone. CONCLUSIONS: These results provide evidence as a proof of principle that smartphones could be a useful tool to objectively assess motor symptoms in PD in clinical and experimental settings.

31 citations


"From wellness to medical diagnostic..." refers background or methods in this paper

  • ...In [3] tests were administered by an experienced clinician in the lab using an HTC Desire device and the collected sensor data were extracted and processed using standard biomedical data analysis software....

    [...]

  • ...Further pursuing this insight, in [3] we investigated the possibility to precisely quantify and implement the mds-updrs methodology as a smartphone app to enable the assessment of motor performance through tremor, gait and bradykinesia measurements obtained from standard sensors embedded in smartphones within a clinical setting....

    [...]

  • ...In CloudUPDRS we employ the data collection and analysis techniques described in [3] to develop an app with extended functionality that enables its independent but dependable use by PwP and their carers at home and in their communities....

    [...]

  • ...As discussed in the previous Section, in [3] we demonstrate the feasibility of using smartphones as a means to assess commonly occurring motor symptoms of PD in a clinical setting....

    [...]