Technology to predict a risky situation at home
The reality in Spain is that 42% of the population wants to stay at home instead of going to a residence or living with a relative. A third of those over 65 fall at least once a year and 50% of them develop a serie of symptoms after the fall which reduces their quality of life. Both things make more and more elderly people die at homes without anyone knowing until several days later.
Nowadays, ageing is a very important issue in the industrialized countries, especially for Portugal and Spain, where the growth trend of this population is high; this information is quite supported by the innumerable existing studies, both those carried out by the United Nations and by the European Commission. Therefore, the union of sociologists, engineers and legal specialists, will allow the integration of technological systems with a high performance of innovative character because there is nothing similar in the market, to provide this way an improvement in quality of life and independence of the elderly, thus decreasing the mortality rate due to lack of early medical attention.
Fortunately, new technologies come in support of the elderly. That is the goal of our project, to create devices that by means of artificial intelligence techniques are able to detect without intermediation the elderly person that has been able to suffer an accident or incident.
In the project we work four universities, two Portuguese and two Spanish and two systems that will monitor the behavior of people and may detect changes in their behavior that may mean that the elderly person is in danger by launching a series of alerts to avoid more serious evils, both systems will be made through the internet technology of things invisible to the person so that in this way enjoy complete freedom. To validate them, a field study will be carried out with both systems and, also, since personal data are being handled, a study will be made about the legal implications that the use of these systems can have.
The aim is to implement a technological system that collects data in real time of the people in your home, and for this it is necessary to bring together the different areas and disciplines that allow the development, field validation and legal analysis of the possible implications of the system. The possibility of bringing together different research groups from Portugal and Spain will allow a better correspondence with the European environment and in this way carry out a sociological and legal analysis that will allow later a greater exploitation of the developed system.
The different research groups have proven previous experience in working with older people and have worked together on various actions, which allows for a closer approach when it comes to solving problems, as this prior cooperation eliminates the barriers that may exist at the beginning of multidisciplinary work.
There will be more than 20 researchers who will work on this project that will allow us to predict a situation of risk in the home. Prediction models have generated enormous progress in artificial intelligence applications (e.g. voice recognition, language translation and driverless vehicles). Applications of similar methods in the social sciences often do not adhere to common standards of information and evaluation, so progress is impossible to evaluate. The reason for this inconsistency is that predictive outcomes depend on many of the same "degrees of researcher freedom" that lead to false positives in traditional hypothesis tests. Depending on which specific group of choices the researcher makes, can be obtained responses that appear to be very different.
To identify the different activities, learning techniques will be combined with invariant and slowly varying characteristics in order to learn hierarchical representations. Specifically, the use of artificial neuron networks in a two-layer structure with three-dimensional convolution and maximum pooling is proposed to make the method scalable to large inputs. An algorithm based on deep learning will be developed for the recognition of human activity by means of sequences of the different sensors. It is therefore a project with a high innovative potential as there is no similar solution on the market.
The Coordinated Programme is composed of four Individual Projects, each led by a university:
1. DSH (Sensory Device for Households). Universidad Carlos III of Madrid (ES)
2. Mobile application in charge of domestic interaction by means of television. Universidad of Aveiro (PT)
3. Field study in Portugal and Spain. Universidade Nova de Lisboa (PT)
4. Analyse the new European legislation on data protection to ensure compliance with the information collected. Universidad of Vigo (ES).
