The Speech Interactive Research Group of the UPV/EHU's Faculty of Science and Technology is leading the European Horizon 2020 EMPATHIC project, which has been rated by European assessors as having a significant social impact. This project is seeking to develop a virtual assistant or coach for senior citizens. Apart from reminding senior citizens about the tasks and activities they can do, this assistant will be capable of picking up mood changes in people through changes in tone of voice or facial expression and will be able to act accordingly in a totally automatic way.
Virtual coach to prolong the independence and healthy habits of senior citizens
The UPV/EHU-University of the Basque Country is coordinating an ambitious European project in which ten international partners are participating
First publication date: 16/10/2017
To keep senior citizens independent and leading healthy lives for as long as possible. This is the social challenge presented by the European Union and which an international consortium, coordinated by the Speech Interactive Research Group of the Department of Electronics and Electricity of the UPV/EHU's Faculty of Science and Technology, has been keen to address. Its proposal consists of designing and developing "a health assistant or coach that is fully autonomous, that interacts with elderly people, and which helps them to go on leading independent lives in the three fundamental mainstays: food, social life and physical exercise," explained María Inés Torres, who is leading the research group and the project.
The final product they want to come up with will be "an avatar that works on tablets, TVs, computers, mobile phones, etc., which by means of sensors, microphones and cameras can hold a dialogue with the users," explained Torres. People will have someone who will speak to them, who will remind them about things, who will suggest activities, etc. For example, if this person has a son or daughter in London, the system will know whether his/her son/daughter is connected to Skype and can suggest giving him/her a call. So it is a system geared towards "elderly but not dependent people; we are targeting people aged about 65 and whose independence can be prolonged".
But beyond functioning as a reminder, the interaction will also allow the system to spot changes in people's behaviour as well as mood changes by picking up changes in their tone of voice, expression and even in their eyes. "Once it has perceived these changes, it will respond in the same way that a coach would, and through dialogue try and correct that situation," specified the researcher. Should the system detect for example that the person has not been out recently, it could suggest an activity that he/she enjoys such as going to a concert, and suggest calling a friend and getting tickets in order to do this.
The importance of bringing positions closer together
The aim set "is hugely ambitious", admits Torres, owing to the large number of disciplines that have to work together, and the challenge posed by "bringing positions closer together and having all of them speaking the same language". The conjunction of disciplines that will need to participate includes professionals of medicine and psychology, all kinds of engineers and technologists, as well as the various groups of end users, on whom one will be relying right from the start to test out and validate everything in which progress is made. "One of the biggest difficulties involves translating the health aims into mathematical models to enable the technology to work." The partners tackling this task include Osatek professionals, who will be providing experience acquired across 6 years of managing the betiON Public Telecare Service, and professionals of Osakidetza (the Basque Autonomous Community public health service) who will be providing knowledge about the health needs of senior citizens.
Then, on the technological side, technologies and systems to address all the functions they want the assistant to have will need to be developed: the avatar that will be interacting with people, automatic speech recognition, voice synthesis and dialogue systems, the system for spotting mood changes through voice signals, facial recognition, the spotting of mood changes in facial expression, the spotting of mood changes in the eyes and the task to translate the health aims into mathematical models to make the technology work.
Finally, three groups of volunteers towards whom the product is geared will be set up so that the progress can be evaluated: one comprising senior citizens who will be attending different types of courses, another of users of telecare services, and a third one of patients selected by medical practitioners.
These groups will be drawn from users of and collaborators in the various Osakidetza-Osatek services as well as from a French association of senior citizens and from a selection made by the University of Oslo. They will all be following similar selection criteria so that they can test the system in different language and cultural contexts.
The consortium set up to run the EMPATHIC project brings together ten European bodies: the above-mentioned Speech Interactive Research Group of the UPV/EHU's Department of Electronics and Electricity (project coordinator), Osakidetza professionals, and professionals from Osatek, which runs the BetiON Public Telecare Service; the University of Oslo; the Mines-Télécom institute of France, specialists in facial recognition; the Technion Institute of Israel, which has neuroscientific knowledge capable of spotting mood changes in the eye; the University of Naples, which has a mixed group comprising people from computing sciences working in the department of Psychology; the London-based firm Intelligent Voice, specialists in speech recognition; the Belgian company Acapela, devoted to voice synthesis; Tunstall, a multinational company with its headquarters in Britain and provider of telecare products and services; and, finally, the French E-Seniors association, set up to tackle the digital divide among elderly people.