#	Pagina
attuale pagina	/open-h2020/projects/199902/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 2 - STARR (Decision SupporT and self-mAnagement system for stRoke survivoRs)

Teaser

Summary

Stroke is a leading cause of death and disability, with an estimated total cost of approximately â‚¬64 billion per year in Europe.
Recurrent stroke carries with it a greater risk than first-ever stroke for death and disability. In the same time, secondary stroke prevention has proved not very successful in the general population. One of the main reasons for these poor results is the fact that quality healthcare outcomes depend upon patients\' adherence to recommended treatments. This adherence remains a challenge, since people do not always understand or remember well enough what they are supposed to do to follow the treatment or to improve their general health status. Furthermore, they do not feel actively involved in a collaborative decision-making process with their physician(s). On the other end of the patient-healthcare professionals relationship, healthcare professionals need to have an understanding of why, how, and when patients do not engage in optimal self-management behaviours in order to engage in a fruitful collaboration with their patients and co-manage more efficiently a personâ€™s health condition.
Thus, better results in the prevention of stroke could be achieved if we improved patientsâ€™ adherence to treatments, the management and self-management of stroke risk factors (e.g. high blood pressure, unhealthy diet, alcohol consumption, physical inactivity) and the collaboration between patients and healthcare professionals. This is the main objective of the STARR project. We developed a modular, affordable, and easy-to-use system, which informs stroke survivors about the relation between their daily activities (e.g., medication intake, physical and cognitive exercise, diet, social contacts) and the risk of having a secondary stroke. The STARR system is based on an existing computational predictive model of stroke risk factors; a number of connected objects integrating off-the-shelf sensors for real-time sensing of proprioceptive functions and simple movements; a vision-based sensing platform for measuring the execution of more complex rehabilitation tasks, as well as for evaluating the stroke survivorâ€™s emotional state; a Decision-Support System (DSS) integrating and processing all this information, evaluating progress towards the achievement of given rehabilitation and lifestyle change goals, and providing the basis for personalised diagnosis and prognosis of the stoke survivorâ€™s health status and of a secondary stroke; a number of cloud services assuring the relations with informal and formal carers, peers and medical staff; a processing unit collecting and distributing the information from the sensors to the different modules; self-management services for stroke survivors giving recommendations and support for improving the adherence to prescribed treatments and adopting a healthier lifestyle.

Work performed

We collected future users\' and stakeholders\' needs, wishes and fears at 4 locations in Europe (France, Spain, Sweden and the UK). In total, 141 stroke suvivors and more than 40 representatives of other stakeholders were involved in the studies. We produced substantial information to design the STARR platform. The system is currently being intergrated and iterative usability tests are done. It will be evaluated in a pilot starting in February 2018.

We have also done a critical analysis of the existing clinical literature on models of stroke risk. We chose 2 models as applicable for the design of STARR (Zuum and the Stroke Riskometer) on the basis of the objectives of the project. Zuum was implemented in the DSS. Also, one model for motion analysis and guidance using a 3D skeleton representation was implemented. It is used for guiding a patient in correctly performing an action or movement by presenting him/her with feedback that is easy to interpret. This model for motion analysis, together with another one on emotion analysis, is currently integrated with the STARR system. We also started developing physoclogical models for behavioural change to halep sustaining patients\' motivation to adopt a healthier lifestyle.
As for sensing, we chose easy-to-install, reliable, low-cost and energy- and performance-efficient sensors (e.g. Kinect Version 2 for vision-based tasks; compact inertial sensors cheaper than 2â‚¬ with a battery lasting one week). The inertial and pressure sensors were integrated inan insole for unobtrusive motion analysis.
Finally, an analysis of the data flows from a data protection perspective was done. The purpose was to ensure that the data flows, which underlie the STARR architecture, comply with the EU data protection legal provisions. Attention was paid that the stroke survivor retains control over the processing of his data and is thus able to see, control and decide what it is they want to share with anyone else.

Final results

STARR is scoping the extent to which existing risk factor management, modelling and technology could be applied to secondary stroke prevention. The originality of the approach lies in the work on the extension of existing models to support independent decision-making by the stroke survivor. Also, all these different models are enriched with real-time data on stroke survivorsâ€™ daily activities. Another original aspect of the project is the feedback given to the user when sensing their motions. This feedback and the associated algorithms for vision sensing have been described in a number of peer-reviewed publications.

A highly scalable DSS has been developped. The original features of the DSS implementation and validation are related to two key areas: (1) PaaS layer for the predictive model execution, which provides easy adaptation and upgrade of predictive models; (2) scalability, allowing the support for growing number of users by means of straightforward (re)configuration. The PaaS layer, which provides the execution environment for the predictive models, is a novel solution, facilitating the split between the predictive model design and validation, and its technical implementation.

A number of self-management services (i.e. mobile applications and games, serious games) are currently being developped. They allow stroke survivors to adopt a healthier lifetyle and practice their rehabilitation alone or with a therapist. Their originality lies in the fact that they are supported by psychological models for health behaviour change.

The general impact of the STARR system on stroke consequences will mainly be evaluated during this pilot. Currently, our impact actions are mainly focused on dissemination of the current project resutls to relevant stakeholders.