Community Research and Development Information Service - CORDIS

FP7

DIPAR Report Summary

Project ID: 262291
Funded under: FP7-SME
Country: Spain

Final Report Summary - DIPAR (Diagnosing Parkinson’s Disease by neuromuscular function evaluation)

Executive Summary:
In the DiPAR project, four European SMEs and four RTDs have developed a digital sensor pen that records biomarkers to aid in the differential diagnosis and monitoring of Parkinson’s Disease (PD).
These novel biomarkers are features, derived from minute motions recorded with movement related sensors. The pen uses algorithms for drawing and handwriting motion analyses to quantify fine motor skill, represented by motion features. These novel biomarkers provide objective information to clinicians.
The pen helps to non-invasively determine if an individual is suffering from Parkinson’s or another motor disorder. The pen mimics the diagnostic assessment that is normally performed by specialists and reveals some of the patient’s motor control mechanisms. The pen will also assist in the continued monitoring of patients.
The device may be used by non-specialists with minimal training to enable decentralized healthcare and screening to improve the capacity of effective and early diagnosis and intervention by neurology clinics to help reduce the overload on European healthcare systems.
The unique pen technology and software is sensitive to major neurological signs of PD and has shown compelling results during the clinical exploratory trials during the DiPAR project. PD patients could be separated from patients with other motor disorders with accuracy of 80% and from healthy elderly with an accuracy of 90%.
The four European SMEs identified five years ago through Newcastle-based medical technology company Manus Neurodynamica Ltd (Manus), an opportunity to develop this brand new disruptive technology for deployment within the clinical assessment market sector. The DiPAR project was funded by the EU’s Research Executive Agency under the R4SME scheme and which was successfully completed in September 2014.
The total number of PD patients was already estimated at 1.2 million in 2005 (0.5% of people over 60 years of age and 2% of people over 80 suffering from PD) and this number is forecast to double by 2030. The current total costs in Europe for PD care is estimated at €13.9 billion.
Although there is currently no cure, there are neuroprotective therapies in development and an early intervention will significantly improve patient care in the future, as protective and most likely regenerative treatment becomes available within the next 5 years.
Before the device will be taken to market, the DiPAR pre-production prototype will be developed into a final product and the pen will embark upon its final stages of testing in, using gold standard DaTScan imaging technology.

Project Context and Objectives:
Project objectives

The SME instigators of the DiPAR project, identified in 2009 through partner Manus Neurodynamica Ltd an opportunity to develop a brand new disruptive technology for deployment within the clinical assessment market sector for early diagnosis of Parkinson’s Disease (PD). Based upon Manus’ existing technology, the company created a project bid to develop:
• A novel, intelligent decision support system, trained against live patient data capable of discerning markers of PD
• A communications framework, supporting secure data transfer between remote clinics, surgeries and healthcare centres
• An enhanced version of the unique and patented hand-held stylus, capable of detecting the minute movements indicative of Parkinson’s, and other neuromotor diseases
• A preclinical trial, conducted at the University Medical Centre Groningen in the Netherlands, demonstrating the efficacy of the system.
Exploitation of the results of the project will lead to improvement of diagnostic healthcare through the DiPAR concept.
The aim of the DiPAR project was to produce a novel and unique set of equipment supporting the diagnosis and rehabilitation of neuromotor diseases. The preliminary proof-of-principle work in this field had been lead by SME partner Manus, who had reached the point where further work, drawing on wider expertise than is at their sole disposal, is required. Recognizing the benefits of collaborative and objective driven research, Manus assembled a vertical supply chain of high-tech SME collaborators, backed by world-leading research base partners to undertake the necessary research and development activities on behalf of the SME partners. Collectively, the DiPAR consortium believed that the optimum strategy to secure the SME-centric involvement in this new technology, and to deliver and commercially exploit the results of this project, was through the funding offered by the Research for the Benefit of SMEs instrument.

The first concept development of the hand held device focussed on diagnosing and monitoring of PD and Parkinsonism. The instigators envisaged that specific neuromotor deterioration associated with PD/Parkinsonism could be assessed by means of externally measuring a set of biomechanical parameters or motion features. Those specific changes that take place within the nervous system with the development of the pathology were known to some extent. However, in order to establish the proposed system, the knowledge firstly had to be extended through clinical research and the development of novel signal processing techniques that could be incorporated into a decision support system. The system development for diagnosis of PD/Parkinsonism could then be completed.

There is enormous scope for establishing the system for diagnosis of PD. First, within the UK and other member states of the EU27, many patients with suspected PD do not have timely access to a neurologist. The novel diagnostic system will enable healthcare workers at different levels of health care, who do not have specialist knowledge of movement disorders, to perform a fast and accurate diagnostic procedure. Secondly, the system will improve accuracy and efficiency of differential diagnosis of movement disorders. In addition, there are indications that early onset PD can be diagnosed before the clinical symptoms have become obvious. For the future, it is anticipated that with extended research, diagnosis of other neurological impairments will become possible and the DiPAR project provides a platform from which further R&D on neuromotor impairments will take place. Another application with significant scope is applying the technology within a training and rehabilitation device for a wide range of impairments or general motor skill improvement. This is further detailed in the section below titled: ‘Other applications: training/rehabilitation and self diagnosis’.

The technological and scientific problem with diagnosing Parkinson’s

Parkinsonism is primarily a disease of the elderly and middle-aged, but can occur in all age groups.
Unfortunately, signs of Parkinson's disease are often still wrongly diagnosed by ascribing the signs of the disability to normal aging. Particularly younger people who have developed a mild initial form of the disease are often wrongly diagnosed as there currently are no appropriate diagnostic methods or products available to clinicians.
A patient can be diagnosed as having a Parkinsonian syndrome or Parkinsonism, if two out of three cardinal signs are seen. These signs are: 1) rigidity (muscular stiffness throughout the range of passive movement in a limb segment); 2) bradykinesia/akinesia (slow/no movement execution), which is most disabling and 3) tremor (postural/during activity) . Not all Parkinsonism originates from idiopathic Parkinson’s disease with an underlying dopamine defect, but could be caused by other pathologies, such as PSP (Progressive Supranuclear Palsy), MSA (Multi System Atrophy) and rarer CBGD (Cortico Basal Ganglionic Degeneration). The usefulness of the diagnostic system will be on diagnosis of Parkinsonism in general, independent of the underlying pathology.

Current diagnostic methods are based on the clinician's subjective interpretation of the patient's performance of writing, spiral drawing and daily living tasks, such as holding a cup, but this interpretation is not objective. Therefore, no reliable comparison can be made from interpretations of samples that were taken at different moments. In addition, the three prominent signs can often not be seen in early Parkinsonism, but 7-10 years before obvious manifestation of the disease, pre-diagnosis neurodegeneration has already set in3. Because of this, diagnosis is not possible until the disease has advanced to a highly significant degree; typically, severe irreversible 60% degeneration of the nigrostriatal neurons has already taken place4, resulting in severe deterioration of motor function and development of non-motor symptoms. The key to early diagnosis, leading to early intervention and an improved overall outcome for the patient lies in measuring and quantifying the earliest changes in the neuromuscular system due to Parkinsonism .
Early diagnosis of Parkinsonism becomes particularly useful with the development of neuroprotective therapy and regeneration . Currently, medication and training allows the symptoms of PD to be suppressed, but unfortunately, no cure exists today. However, treatments that enable neuroprotection and regeneration are likely to become available in the near future. When these neuroprotection and regeneration treatments become widely available, preventive screening for PD before manifestation of symptoms will become a matter of urgency.
Although no non-invasive automated diagnostic tools for Parkinsonism exist anywhere within the EU, it must be stressed that the benefits and the usefulness of such a system differs between the various EU member states. In the UK, for example, the usefulness of the system lies in diagnosis of Parkinsonism as a shortage of neurologists leads to waiting times for investigation by a specialist up to 52 weeks and a late diagnosis is common in the UK.
Currently, 0.5% of people over 60 years of age and 2% of people over 80 years suffer from PD.
There currently are less than 400 neurologists in the UK and the ratio of neurologist: patients is currently 1:177,000, compared with 1:26,000 (US) and 1:8,000 (IT); see figure 1. According to the Association of British Neurologists (ABN), the minimal ideal ratio would be 1:43,000 .
The ABN state: "Neurologists want to ensure that people with neurological disorders have timely access to a high quality, comprehensive and coordinated, patient-centred expert service with equity of provision regardless of age, race and gender." . In addition, pressure to see outpatients to meet NHS targets has reduced the capacity to follow up many patients with chronic neurological disorders, who require long term specialist care if secondary complications are to be avoided . Consequently, patients often do not get the required care 12 13. It can be concluded it is highly important that the neurologist's time is spent efficiently.
The proposed system for fast and reliable diagnosis and disease monitoring enables us to facilitate the required improvement of neurology health care. The diagnostic system for Parkinsonism will enable general practitioners, nurses and others not specialized in neurology to perform the required diagnostic processes by means of the automated diagnostic tool. In the UK, 10.000 people are diagnosed every year with Parkinson's disease. With the trend of decentralized health care in the UK, the novel diagnostic system enables health care workers to timely assess all potential cases of Parkinsonism and start treatment without taking up more time of the specialist than required.
Consequently, immense healthcare costs could be saved every year.

Figure 1: Distribution of neurologists in various European countries: population per neurologist.
However, in EU member states where patients have timely access to neurologists, all required diagnostic tests can also be performed by specialists. In those cases, the usefulness of the system will firstly be improved differentiation between different neuromotor impairments, e.g. between PD and other forms of tremor by means of improved quantification of symptoms.

In addition, it has been recognised that PD patients are often still wrongly diagnosed. Diagnostic error rates of 47% 18) and 26% were found . A lower error rate was reported in standard neurologic and geriatric practice and special movement disorder clinics 18 . This indicates that expert knowledge of movement disorders enables a more accurate differential diagnosis, which is supported by the fact that the diagnostic error rate appears lower in other parts of the world, where more clinicians are available. A study by Jankovic et al (2000) in the US reports an error percentage of 8.1% . However, in 2008 it was found from a study in 10 European sites that PD is still commonly misdiagnosed in Europe .

In order to lower the diagnostic error rate, developments of accurate diagnostic tests for PD are of high and immediate importance. The novel diagnostic tool enables objective assessment of neuromuscular processing involved in writing by means of comparing a set of kinematic and kinetic data parameters with a database containing upper and lower limits to ranges of those parameters.

Several EU consultant neurologists also revealed during interviews that they anticipate a need for the novel technology in preventive screening. Preventive screening will become common practice within the next 5-7 years once the neuroprotective and neuroregenerative therapies, which are currently in development, have become established. In addition, it is anticipated that further discrimination between a number of other neurological disorders than PD and diagnosis of those disorders becomes feasible at a later stage.

The system for diagnostics of Parkinsonism will be targeted at physicians and hospitals, who will be approached both directly and through the private and national health services. It is anticipated that facilitating each neurology department with the proposed system for fast and reliable diagnosis and disease monitoring could significantly improve the efficiency of patient care in neurology, which will also lead to improved economic efficiency. An elaboration follows here.
Diagnosis is the key to making an accurate judgement on the most appropriate medical or surgical treatment 4 . Monitoring the disease is required for further management of the disease and re-diagnosis is required 4 with the development of atypical symptoms. A late or wrong diagnosis will certainly lead to a progression of the disease. At that stage secondary complications and extensive costs for daily care, physio- and occupational therapy, treatment with medication and sometimes surgery cannot be avoided anymore4. Although the current trend for some forms of Parkinsonism is to not immediately start treatment with medication3, research results suggest that early diagnosis and treatment is required to improve the patient’s health and quality of life4. In all cases, early diagnosis will at least enable to start physical therapy and will lead to better understanding of the disease and response to treatment.

Project Results:
Overview

Four European SMEs and four RTDs have collaborated to deliver a pre-production prototype of the system to aid in the differential diagnosis and monitoring of Parkinson’s Disease. The project scientific and technology results include new hard and software and clinical exploratory trials results.
The diagnostic system consists of a pen-like hand held device with a novel sensor system for recording biomarkers. The software uses algorithms that derive features from minute motions recorded with movement related sensors that enable to quantify fine motor skill. The decision support system, integrated with the software, enables to provides objective information to clinicians.
The final system has been taken through pre-clinical trials, conducted at the University Medical Centre Groningen (Netherlands) to demonstrate the efficacy of the system. The decision support system was developed by the Technical Research Centre of Finland with support from the German SME Pattern Expert.
Whilst the technological advancement of sensor and computer technology has allowed for simple assessments of the nervous system, there are huge technological challenges associated with developing a device that can accurately collect data and catalogue symptoms of such diseases. The majority of the technical challenges have been resolved in the DiPAR project. The result is a pre-production prototype of a unique piece of medical technology that links ten years experimentation with sensor systems and developing data analysis methods with the key features of Parkinson’s. The (prototype) sensory pen has been patented.
The project scientific and technology results will be described for each of the four main project objectives: Developing the hand-held stylus; Developing the decision support system; Developing secure patient data transfer platform; Conducting a clinical exploratory trial.

Developing the hand-held stylus

Delivering an enhanced version of the unique hand-held device involved collaboration between the Fraunhofer Institutes IPA (Stuttgart) and IPMS (Dresden), the German SME inotec and UK based SME Manus, the project instigator.

The overall system consists of a target system (tablet computer), the sensor pen and the wireless charging station. The sensor system integrated with the stylus is a unique chipset that is able to capture accelerations, angular velocities and forces acting on the sensor pen. The collected sensor data is transmitted via a wireless connection from the sensor pen to the USB wireless adapter. The host software on the target system (Windows or Linux computer) can access the measurement data and communicate with the sensor pen via a “Virtual COM Port”. To avoid disturbances due to power cables, a rechargeable battery is used as power supply. Thanks to the wireless battery charging station, the housing of the pen is sealed for ease disinfecting.

The system performance requirements and operational standards were derived in collaboration with the clinical partner UMCG and clinical data analysis experts at VTT and Manus, the commercial entity that provided the market access a report. The partners discussed their findings upon which conclusions were drawn that were subsequently evaluated by external advisors before using them as input to the hardware design.

Previous issues with off-the-shelf tactile force measurement sensors (such as inaccuracy due to drift, temperature effects and bulky size) were overcome in a joint effort between SME partners Inotec (Leipzig, Germany) and Manus (Newcastle, UK) to implement miniature strain gauge sensors. This resulted in increased sensor accuracy and successful stylus miniaturization. The solution provides high accuracy and high resolution force recordings that can be integrated with the pen without compromising on handle ability.

An upper limb movement detector was required as input to the diagnostic sensor system and this was provided by Fraunhofer IPA. The upper limb detection system records motion of hand, lower arm, upper arm and shoulder of the patient. An integrated 6 DOF IMU device with gyroscopes and accelerometers was implemented. Synchronization between limb detection system and pen was implemented by means of an optical trigger.
This final pre-production prototype pen was completed in March 2014 for use in the clinical validation in the project.

Developing a decision support system

The second major project objective was developing a novel, intelligent decision support system, trained against live patient data capable of discerning markers of PD. The mathematical algorithms that deliver these biomarkers with diagnostic and monitoring capability were developed by Dr. Mark van Gils and colleagues at the Technical Research Centre of Finland (VTT) with support from the German bioinformatics SME Pattern Expert.

More specifically, the develop signal processing techniques should allow to extract features from the biomechanical signal sensors in the pen-based device that allows classification between:
(1) Data from PD patients and healthy subjects;
(2) Data from subjects having different types of tremor pathology, and
(3) Data from patients with different types of muscle control deficiencies that cause movement in-balances.
Classification performance in terms of sensitivity and specificity was to be (as a minimum) comparable to current diagnosis practices. Furthermore, the signal processing methods and feature extraction methods took into account hardware-limitations of the system, inter- and intra-patient variability in the signals, artifact detection, and feature robustness.
The work was part of an exploratory clinical validation, and carried out in close co-operation with WP3 – that provided input in the form of data collected from the studies, as well as expertise in the interpretation of the data. To reach the goals, the work was divided into 7 tasks that were worked on throughout the project, which included:
• Task 4.1 Development of methods for signal validation, artefact detection and rejection
• Task 4.2 Development of methods for signal quality improvement
• Task 4.3 Development of methods for feature extraction
• Task 4.4 Feature selection
• Task 4.5 Classification
• Task 4.6 Performance Evaluation
• Task 4.7 Delivery of scripts to WP3 and WP5

The set of tasks (Table 1 in the publishable summary pdf) that participants performed using the DiPAR hardware evolved over the duration of the project (also see the WP3 section).

From the data recorded during these tasks many different features were extracted (Table 2). Feature sets evolved throughout the period – some were found to be more suited than others, and some ‘scientifically interesting’ features were not pursued further in the cases where their contribution to classification performance was not significant.

The goal of work package 4 was to calculate features that can subsequently be used as inputs for a classifier. To evaluate the performance of the features we also did some classification analyses, initially using simple binary one-feature classifiers to analyse the performance of individual features, ranking the features by AUC-ROC.
The first deliverable reported 193 distinct features. Testing the performance of individual features for the discrimination problem of separating healthy controls from Parkinson’s disease patients showed that 48 of these features are statistically significant at the 5% level and 16 of these 48 are statistically significant at the 1% level.
The second deliverable reported on 624 features, and evaluated their performance on the discrimination problem of separating Parkinson’s disease patients from the other participant groups studied in that deliverable (healthy controls, essential tremor patients, functional tremor patients and enhanced physiological tremor patients). Table 3 in the publishable summary pdf summarizes the performance of features calculated by the various feature calculation methods from D4.2.
There are 21 features that by themselves satisfy the pre-set performance requirement of sensitivity >
0.8 and specificity > 0.8.
While most performance evaluation was done using simple per-feature binary classifiers, more multi-input classifications were also explored.
In summary, the separation of the PD patients from healthy controls seems to work well; except for the oldest group of healthy controls, in which the classification accuracy is lower than what was obtained using the study 1 data. One possible cause for this is higher variability in the PD group; patients were for example measured both on and off medication while in study 1 all the patients had a break in the medication.
The separation of the different tremors looks also promising; the essential tremor was the easiest to separate from PD with classification accuracies around 80 %. The enhanced physiologic tremor was most difficult to separate from the PD. The classification accuracy could possibly be improved by redesigning the tasks so that the rest task would clearly measure rest tremor.
The medication seemed to have the expected effect on the classifier outputs, the patients looked more like healthy controls when they were on medication, but the results can also be the result of a learning effect. More detailed analysis of the changes between the on medication and off medication conditions should be performed to find changes that are unlikely to be caused by the subject learning the tasks.
Overall the classification tests give promising results, but further analysis is needed. The exact clinically relevant scenarios in which the DiPAR system would be used should be defined and the populations in the classification tasks fixed accordingly. Also the number of features used by the classifiers should be reduced; this would lead to more human understandable decision making and also quite likely to better classification accuracy.
Developing secure patient data transfer platform
Crucial for smooth implementation was the third deliverable: Developing a communications framework, supporting secure data transfer between remote clinics, surgeries and healthcare centres.
This communications frameworkhas been developed by Glasgow University, assisted by Spanish IT company Hispafuentes.

The system has been used for collection and security-oriented transmission of all data sets between the diagnostic systems and the centralized database currently hosted at UGlas. The database is accessible to DiPAR consortium members to train and refine their algorithms to detect movement disorders; to support the clinical trials process, and for future research into movement disorders.
Security-oriented access and usage policies with fine grained security will be used to ensure only trusted and authorised individuals are able to access associated data sets.
The live system, contains participant data from all 4 available studies uploaded and the system is fully documented (see appendices). Due to the closure of the National e-Science Centre, UGlas was further tasked with updating the system for future development, and relocating the system to run beyond the life of the DiPAR project to support future trials and commercial implementation.
The final hard and software solution that was implemented by the University of Glasgow comprises the following:
# Original source code for Data Collection Platform
# Authored in Java, utilising the Vaadin environment and Liferay user management classes
# Fully configured and operational DiPAR Liferay web portal, with collaboration tools, public facing content pages and pre-installed Platform tool
# https://dipar.nesc.gla.ac.uk
# MySQL Database for hosting DiPAR data
# User guide for Platform (in Appendix WP6-1)
# Manual for non-admin users of the platform, with screenshots
# Administrator guide for Platform (with database schema) (in Appendix WP6-2)
# Manual for portal administrators and installers describing the installation, configuration and administration process, plus build instructions for the Platform tool
# Standard Operating Procedure (SOP) Documents
# Confidentiality Policy Document v1.1
# System Level Security Policy Document v1.1
# Two high-spec machines for running infrastructure
# one for internet-facing portal, other for privately networked database
Conducting a clinical exploratory trial

The final project objective was conducting a clinical exploratory trial, demonstrating the efficacy of the system and the system has been taken through pre-clinical trials at the University Medical Centre Groningen (Netherlands). The trial enabled to test the decision support system that was developed by the Technical Research Centre of Finland with support from the German SME Pattern Expert (section 1.3.4).

The clinical validation involved data collection with the DiPAR system, analyzing the data for relevant patient groups (in close collaboration with VTT), to allow the comparison of movement features parameters between these groups. In addition, patients with Parkinson’s disease (PD) were assessed on and off medication (as part of study III), a repeatability study was performed in healthy subjects and a prospective study was performed in an unselected group of patients with movement disorders (including PD) visiting outpatient clinics in tertiary centers. The exploratory trials thus entailed five sub-studies:
I Comparison between PD patients and healthy control participants (executed with Manus pen prototype (pen V2011))
II Comparison between patients with different types of action tremors (essential tremor (ET), enhanced physiological tremor (EPT), functional tremor (FT)) (executed with DiPAR pen prototype 1 (pen V2012))
III Comparison between groups of patients with movement disorders that involve an imbalance in muscle coordination (PD, writer’s cramp, spasticity), including an assessment in PD patients both on and off medication to assess the influence of medication on DiPAR test performance (executed with DiPAR pen prototype 2 (pen V2014))
IV Repeatability study: Assessment of healthy participants of different ages at two occasions with one week in between (executed with DiPAR pen prototype 1 (pen V2012)).
V Prospective study: Assessment of unselected patients visiting the Movement Disorders outpatient clinic of a tertiary referral center (UMCG and the Neurological Institute of the Mater Misericordiae University Hospital in Dublin (Ireland)). The results were used to classify patients as having PD or other movement disorders (including other Parkinsonisms), based on the definite diagnosis as obtained later from the patient file, employing the analysis methods developed for studies I-III (executed with DiPAR pen prototype 1 (pen V2012)).

At the end of the project, it was concluded that the complete system (pen with tablet, standardised tasks and analysis software) will be useful in clinical settings to support diagnostic assessment, treatment evaluation and patient monitoring. The system provides objective measures to characterise motor symptoms of movement disorders, such as Parkinson’s disease (PD) and different types of tremor disorders:
• The standardized tasks have high reproducibility, similar to that of the Purdue pegboard task, a reference measure for evaluating fine motor control.
• The sensory pen allows to distinguish PD patients from age-matched healthy controls with a sensitivity and specificity of 90%.
• Several quantitative features derived from the handwriting and tracing tasks differ between patients with different types of tremor disorders.
• UPDRS (Unified Parkinson’s Disease Rating Scale) scores (limited to hand function) correlate with several quantitative features, with similar correlations as for the Purdue Pegboard test with these features. This indicates that the standardized tasks, which measure fine motor control, allow to monitor medication effects and disease progression in PD patients as far as they affect upper limb/hand motor performance. Furthermore, treatment effects on the performance on these tasks reflect what is known about treatment effects in PD.
• Several quantitative features derived from the handwriting and tracing tasks differ between PD patients, tremor dominant PD patients and patients with other movement disorders.
Novel analysis methods were delivered by the Technical Research Centre of Finland (VTT). The full discriminative power of the sensory pen system was assessed combining all tasks and features from minute pen motions:
• When separating PD patients from healthy controls, essential tremor patients, functional tremor patients and enhanced physiological tremor patients, we identified 21 features that by themselves satisfy the pre-set performance requirement of sensitivity > 0.8 and specificity > 0.8. The best performing features with the highest AUC-ROC values are all related to drawing speed (AUC 0.84-0.94). The superior features in this category are derived from the time taken to draw ‘e’ or ‘l’ shapes in the ‘elelelel’ task.
• The separation of PD patients from healthy controls seems to work well; except for the oldest group of healthy controls, which was due to the study setup, with no distinction being made between patients being on and off medication. This resulted in the classification accuracy being lower than what was obtained using the study 1 data. In study 1 all the patients had a break in the medication, resulting in a sensitivity and specificity of 90%.
• The separation of patients with different tremors also looks promising; essential tremor was easiest to separate from PD with classification accuracies around 80 %. Enhanced physiological tremor was most difficult to separate from PD. The classification accuracy could possibly be improved by redesigning the tasks so that the rest task would clearly measure rest tremor.
• Overall the classification tests give promising results, but further analysis is needed. The exact clinically relevant scenarios in which the DiPAR system would be used should be defined and the populations in the classification tasks fixed accordingly. We expect that reducing the number of features used by the classifiers will lead to more human understandable decision making and also to further improvement of classification accuracy.

Potential Impact:
Markets and users

Three groups of applications for the technology are envisaged:
A. Clinical diagnosis and management of Parkinson’s Disease
B. Academic and industrial research applications
C. Non-healthcare market applications
The first group of applications (Group A) for this technology comprises a hardware and software platform for diagnosis and management of PD. Manus has identified a number of diagnostic applications within the overall PD care process. In addition, there are opportunities for the Company’s products in the management of patient care and treatment after the initial diagnosis has been made. These include monitoring of disease progression, drug efficacy testing and rehabilitation.
For the future, an opportunity in pre-symptomatic screening of apparently healthy individuals is also envisaged. The commercialization of the ‘Group A applications’ is the primary focus of the Company’s activities.

Group B applications lie in the fields of academic and industrial research, including the support of drug development. In these markets, the same hardware will be used as in the primary (PD) market with only minor modifications to the front end software for specific research needs. This second application group is attractive as it will enable earlier market entry. Moreover, the same technology can be utilised to exploit this market and additional costs to reach this market are relatively small compared to those already on-going for group A.
A number of well established clinicians and scientists have been identified as early adopters for the technology and have agreed to purchase sensory pen system for research purpose as soon as it becomes available. In addition, market research in the pharmaceutical industry has resulted in interest from potential users.

The Group C applications lie in non-medical markets. Currently envisioned markets include: virtual reality training and computer control for disabled people; supporting development of personal care products; rehabilitation and training of handwriting and upper limb skilfulness in both healthy young people and people suffering from neuromotor impairments; and possibly also direct-to-consumer marketing of a low-cost monitoring device for in-home use. Some of these applications require significant additional product development work.
With regard to clinical use (group A products), the applications will imply changes in the ways clinicians manage their patients and their practice groups. Therefore, through conducting market research between March and November 2013, Manus determined first the likely degree of interest in products for these applications, and second, what conditions (for example focused clinical trials) will need to be satisfied in order for this demand to develop. These results have guided the company’s needs to investment in product and market development and resolve barriers to market entry.
Finally, a third group of direct-to-consumer products can be developed within 5 years (biometrics, human-computer interface, training at home etc).

User benefits

Early diagnosis of Parkinsonism is an issue on which there is considerable debate currently among PD specialists. While currently medication and training allows the symptoms of PD to be suppressed, no actual cure exists today. The general consensus in the field of PD research seems to be that early diagnosis is increasingly important for the best possible patient outcome . Moreover, there is near universal agreement that early diagnosis will become essential once neuroprotective or regenerative therapies become available . This is a result of a large number of potentially promising methods for treatment, such as improved drug delivery and control, and training and rehabilitation to maintain or improve dexterity and mobility. Many experts also believe that treatments that enable neuroprotection and regeneration are likely to become available in the near future. Some examples of these include:
(a) Glial cell derived neurotrophic factor (GDNF), a protein that has been shown to help brain cells grow and could be used to grow new dopamine-producing cells to slow the progression or stop Parkinson's disease . Based on this work, UniQure BV has developed a new gene therapy treatment for Parkinson's disease.
(b) Eli Lilly (LLY) and the medical device giant Medtronic (MDT) have announced the formation of a research and development collaboration to tackle Parkinson's disease by delivering a targeted dose of an experimental medicine through an implantable drug delivery system.
(c) A number of other companies and academic groups, including Domain Therapeutics (working on a nano-engineered delivery method for GDNF),and Eli Heldman and colleagues (Lauren Sciences, LLC and Neuroderm) are known to be working on nano-engineered drug delivery systems for PD.
(d) Several groups are also working on novel gene therapy approaches. Oxford Biomedica is a leading example, with their PD gene therapy invention ProSavin with Dr Philip Buttery, from the Cambridge Centre for Brain Repair, leading the British arm. Another such company is Neurologix (working on a gene that is transferred into cells by an inert virus to make a chemical called GABA, a major inhibitory neurotransmitter in the brain that helps "quiet" excessive neuronal firing related to PD .
Studies involving hundreds of patients will be needed to confirm that these therapies are safe and effective and even if they fulfill the promise that they are already showing, it could be five years or more before treatments become widely available. However, when these neuroprotection and/or regeneration treatments reach the market, pre-symptomatic screening for PD before manifestation of symptoms will become a matter of urgency.

Timely access to specialists by patients with suspected symptoms is also an issue in some countries (see also section 1.2). Consequently, patients often do not always get the required post-diagnostic care xi xii. For both initial diagnosis and follow-up care, therefore, it is highly important that the neurologist's time is spent efficiently.
It has been recognised that PD patients are often still wrongly diagnosed. In some studies, diagnostic error rates as high as 47% ) and 26% were found . A lower but nevertheless significant error rate was reported in standard neurologic and geriatric practice and special movement disorder clinics . This indicates that expert knowledge of movement disorders enables a more accurate differential diagnosis, which is supported by the fact that the diagnostic error rate appears lower in parts of the world where a greater number of specialised clinicians are available. However, a study in the US, where specialist to population ratio is relatively high, reports an error percentage that is still quite high, at 8.1% . Also, in 2008 it was found from a study in 10 European sites that PD is still commonly misdiagnosed in Europe . In order to lower the diagnostic error rate, developments of accurate diagnostic tests for PD are of high and immediate importance.
The situation relative to access to specialists is less critical in most of the other countries in the EU where patients have more timely access to neurologists. In those countries, the usefulness of the system will mainly derive from reduced error rate in diagnosis and improved differentiation between different neuromotor impairments, e.g. between PD and ET and other forms of tremor or dystonia.
According to Dr. Duddy, consultant neurologist at the Royal Victoria Infirmary, Newcastle, UK:
“One of the difficulties with diagnosing Parkinson’s diseases in the early stages is that it can be mimicked by other more benign conditions, particularly essential tremor (ET). Many people with ET worry that they have Parkinson’s disease and this causes a lot of concerns for patients and their GPs and generates a lot of work for neurology departments sorting the two out.” Differentiating between Parkinsonism and ET can be difficult even for experienced physicians and diagnostic errors are not uncommon.

Given the relatively high rates of incidence of PD (about 0.5% of people over 60 years of age and 2% of people over 80 years), pre-symptomatic screening of large numbers of elderly people may become very important. Such screening will likely become common practice once the neuroprotective and neuroregenerative therapies are established. But it also has the possibility of becoming more prevalent even before such therapies are available, since professional opinion about the benefits of early diagnosis and treatment, even with existing medications and therapies, varies widely.
In order to diagnose earlier, lower the diagnostic error rate and carry out pre-symptomatic screening, developments of accurate diagnostic tests for PD are of high and immediate importance. The DiPAR project foreground offers a unique solution to this need through objective assessment of neuromuscular processing involved in writing by means of deriving a set of kinematic and kinetic data parameters and interpretation of these features. In addition, further discrimination between neurological disorders and diagnosis of disorders other than PD (e.g. ET) becomes feasible.
To conclude, early diagnosis of PD avoids early deterioration and consequent long-term patient outcomes tend to be better and avoid or delay the need for long-term care, leading to significant cost reduction. Therefore, wide spread adoption of the sensory pen will directly result in a positive economic impact to health care providers.
With regard to use of the pen in research and drug development (Group 2), as stated above, studies involving hundreds of patients will be needed to confirm that new neuroprotective therapies are safe and effective. There are currently 1178 clinical studies on Parkinson’s disease currently underway worldwide, which is a 20% increase compared to a year ago and 360 of these trials takes place in EU and the remainder in US and Canada (approximately 700) .
The sensory pen solution will for the first time enable to quantify patients’ level of motor skill to rate the disease state more accurately than the current standard, the Unified Parkinson Disease Rating Scale, which is quite coarse. More accurate rating means that more significant data can be obtained and that less time is required for one trial, which leads to cost savings on PD (drug) trials. The many positive responses from pharmaceutical companies in EU and US confirm this. Several on-going discussions with clinicians and pharmaceutical companies illustrate how dynamic the landscape is and how the technology may support other research and facilitate the development of new applications.

Summary of wider impact

The economic impact of PD includes: (a) direct cost to the EU health care services under private and government funded schemes; (b) indirect cost to society and financial impact of PD on individuals with the condition and their family and carers.
Direct costs are more easiliy measured. By means of an example, in the UK the direct costs of treatment to the National Health Service (NHS) have been estimated at approximately £2,298 per patient per year. Significant cost drivers include the onset of motor fluctuations and dyskinesias.
Falls are common and often lead to fractures and prolonged periods of hospitalisation. The total annual cost of care including NHS, social services and private expenditure per patient in the UK has been estimated at approximately £5,993 (€9.775). This results in direct costs of approximately £750 million per year in the UK for approximately 127,000 individuals with PD. Costs to the NHS were approximately 38% of the total costs.
According to the European Parkinon’s Disease Association (2012), the total annual direct cost per PD patient for care in Sweden, France, Czech republic were repectively €8.328, €4.421 and €5.510 and for Germany total annual cost were as high as €18,660-31,660, dependent on the phase of the disease. The total costs in Europe to care for 1.2 million patients is estimated at €13.9 billion . Total costs of care increase with age and disease severity .
Indirect costs arise as a consequence of the condition, but are not directly related to clinical disease management. Such costs may include loss of employment for the patient and the costs of additional home or institutionalised care .
Typically 70% of the economic burdon of PD is related to long term disability as also reflected by the above stated increase in costs for care with disease progression in Germany. This may be prevented and delayed with early correct diagnosis and improved treatment. This is a result of a large number of potentially promising methods for treatment, such as improved drug delivery and control , and training and rehabilitation to maintain or improve dexterity and mobility . Many experts also believe that treatments that enable neuroprotection and regeneration are likely to become available in the near future.
Modelling the financial impact that the product could potentially make to reduce costs of care is rather complex and would require accurate knowledge regarding the point in the care process at which it will be used and by whom (e.g. by neurologist or PD nurse, as triage tool, or as confirmation, or alternative to DaTSCAN etc). This is not possible until clinical trial data are available as a result of this project. Nevertheless, a cost comparison with the use of DaTSCAN is likely to be favourable. We estimate that the cost of a 15 minutes assessment assessment with the sensory pen by a PD nurse will be in the order of €70,-, compared to around €1000-1400,- for a DaTSCAN. Given that our system could help to make an early accurate diagnosis that could help early treatment and avoid patient deterioration to the stage where intervention costs increase, huge costs savings may be achieved for EU wide health care providers, individuals and society.

We envisage that less experienced clinicians and non-specialist healthcare workers may carry out initial tests to triage suspected patients with the system for early appointments with the consulting neurologists and so help to reduce waiting lists. Earlier and confirmed clinical diagnosis based on quantification of symptoms will have direct patient benefits from early adoption in 2017 with a step-wise increase of the scale-of-use until 2021, especially with the development of neuroprotective therapies.
The product will push forward the development of medication for treatment of PD as the effects of medication can be objectively quantified for the first time and linked to disease stage. This can take place at thigh time resolution to speed up trials and decrease costs. The current gold standard disease rating scale, the UPDRS, is subjective and quite crude and gives far greater attention to gross motor skill than to fine motor skill.
The product implementation will enable clinicians, for the first time, to monitor the motor aspects of PD patients’ health in long term studies and assess the overall care process from the initial diagnosis onward. This will shed light on the on-going debate among clinicians about the importance of early diagnosis and intervention. It is anticipated that a five year study will be required to inform clinicians, policy makers and commissioners over the course of the study about the advantages of early diagnosis and intervention. When sufficient evidence is obtained, pre-symptomatic screening for PD may be introduced from 2020.
Following on from this, younger, more active patients may be treated in a timely manner and remain active for longer and may they also influence medical policy makers and determine where R&D budgets may be spent, which also has the potential to improve PD care.

List of Websites:
www.dipar.net

Related information

Reported by

Hispafuentes S.L.
Spain
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top