Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - WellCO (Wellbeing and Health Virtual Coach)

Teaser

WellCo provides a virtual coach aimed at fostering the adoption of the healthier behaviour choices in people in order to reduce the impact that bad behaviours in physical activity, nutrition, mental status or social wellbeing usually have in the status of the user in the long...

Summary

WellCo provides a virtual coach aimed at fostering the adoption of the healthier behaviour choices in people in order to reduce the impact that bad behaviours in physical activity, nutrition, mental status or social wellbeing usually have in the status of the user in the long term. The solution proposed leverages data from sensors worn by the users, validated questionnaires and other information about users’ daily life, preferences and context to provide recommendations that ensure the adoption of healthier behaviour choices by users. According to the behaviour change wheel model, the personalization of recommendations assesses the capability and opportunity of the user to perform the change and determines the best way to provide it in order to motive the user in its follow-up. Several functionalities to encourage the user in the achievement of goals and recommendations exist, e.g. social network, supporting groups, etc. These functionalities set the basis of the WellCo Senior App.
Other two applications will be available for WellCo: Informal Caregiver App, for informal caregivers or family, and the Expert Web App, for experts and formal caregivers.
Under this purpose, the following objectives have been set in WellCo:
• Develop novel ICT approaches for useful and effective personalised recommendations and follow up in terms of preserving physical, cognitive, mental and social well-being for as long as possible.
• Validate non-obtrusive technologies for physical, cognitive, social and wellbeing.
• Evidence of user-centred design and innovation, new intuitive ways of human-computer interaction and user acceptance.
• Cost-effective analysis to maximize the quality and length of life in terms of activity and independence for people in need of guidance and care due to age related conditions because of self-care, lifestyle and care management.

Work performed

The project started defining the basis for the proper management of quality and risks in the project was carried out; thus, the first Risk Management Plan as well as different templates for Quality Management were created. All this information is in D1.1 Project Handbook.
In parallel, end-users’ partners were focused on the definition of the user involvement plan (T2.1) where the target users, procedures followed to recruit them as well as to ensure that trials are conducted in compliance with fundamental ethical and data protection principles and legislation was performed. As consequence of this task, D2.1 Engagement Plan and Risk Mitigation Protocol as well as D2.2 Ethics/gender and Data Protection Compliance Protocol (M5) were submitted. Later, partners started the phase of requirements definition in WellCo (T2.2). This phase followed a qualitative methodology using Cultural Probes, diaries and interviews with users that was mixed with quantitative data from an additional questionnaire. These initial requirements are in D2.3 End-User Requirements Report (M7).
After the formulation of the end-users’ requirements, the process to transform end-users’ requirements into concepts for the design of WellCo was started (T2.4) with the use of personas and scenarios to get a clear picture of the end-users’ needs in relation to the already defined requirements. With this information, a low-fidelity visual design was created and that are collected in D2.4 WellCo Design Document and Mock-up. Results were collected in the first iteration of D2.5 Pilots Validation Report together with an updated requirement list(M8).
In M6, WP3 was started with a deep search in the market of wearable devices were the identification of those devices that best matched with the objectives pursued in WellCo. After selecting the devices, an initial design of WellCo architecture was included in D3.1 WellCo Architecture (M13). These ideas served for the implementation of prototype 1 (delivered in M18) and D3.2 WellCo Interfaces and User Manual.
In M7, a study of the main variables, measurements and context that could affect the wellbeing of user was performed in D4.1 User State Assessment Model. The extensive work performed in this work packages can be extracted from the deliverables associated to each module, i.e. D4.3 Nutrition Monitoring, D4.4 Physical Health State Monitoring (M15) and other deliverables expected to be submitted in the coming months.
In M12, initial results, dissemination plan and KPIs for the coming months are collected in D6.2 Plan for Exploitation and Dissemination of Results. Also, an initial document with the initial ideas for open access was submitted in M5 (D6.6 Open Data Management Plan). Finally, as of M12, initial work was performed in individual exploitation with the delivery of D6.4 Exploitation & Business Model for every country where trials are carried out. Special effort has also been put in innovation, where an initial iterative innovation process has been collected in the first release of D6.3 Innovation Management Plan. Finally, initial steps for open innovation and standardization were gathered in D6.5 Standardization and Open-Innovation Report, although they were very brief due to the lack of tangible results at M12.

Final results

• Users’ Assessment: WellCo in addition to monitor physical, psychological and mental activities of the user, it reliably and timely integrate this data with other kind of data coming from heterogeneous sources such as validated questionnaires or users’ profile. In this way enables proper assessment of users’ state and adequate estimation of risks factors that could impact users’ wellbeing in the long term; assuming evidence-based approach along the whole development and deployment cycle.
Impact. This module could be used to support the decision of medical experts when providing recommendations for healthier behaviour to patients.
• Affective Computing via visual features and speech: WellCo incorporates a deep learning classifier that incorporate several convolutional layers aimed at catching the facial features indicating the emotion of the user. Due to the difficulty of categorizing some emotions using only on face visual features, e.g. fear could be mislabelled as surprise and vice versa, WellCo project creates a model that ensembles the outputs from the classifier of visual features and speech at decision level in order to provide more accurate classification.
Impact. Through the affective module, the virtual coach is able to catch users’ emotions and thus, generate the most suitable dialogue and expressions in order to set emotional connection with the user.
• WellCo: Behaviour Change Recommender: Unlike most recommenders, that only considers aspects of behaviour of users within the app, WellCo merges this data with other data coming from sensors, Life Plan and validated questionnaires with the aim of providing adequate knowledge about the user and the context surrounding him. Under this purpose, four recommenders are initially envisaged in WellCo: wellbeing recommender (provides recommendations based on users’ status and wellbeing expectations), emotion (recommendations to change the mood of the user), social (recommendations to foster social interaction in WellCo) and general recommender (tips for wellbeing). These recommendations are automatically adapted to users’ preferences.
Impact. Personalization enables to provide recommendations adapted to the capabilities and opportunities of the user, favouring the motivation and adoption of them by the user.

Website & more info

More info: http://www.wellco-project.eu.