In this paper we present the architecture of a platform with the goal to support social, context-aware and adaptive/persuasive services aimed at stimulating the elderly to stay active/occupied in life. The platform is Webbased and consists of several modules whose main features are presented and discussed in the paper.
INTRODUCTION
Nowadays, population ageing is a common phenomenon in
various countries, which has important implications on the
labour market, as the ratio between the number of inactive
individuals and the ones still active in the workforce will
continue to rise [
In order to address this trend, the approach described in this paper consists in providing the elderly with a novel, technological Web-based context-aware platform offering social features (e.g. sharing knowledge, crowdsourcing services, etc.) aimed to support, motivate and persuade seniors to continue to be active in the workforce so promoting their lifelong workability.
The use of persuasive strategies for encouraging elderly
people to adopt healthy lifestyle habits has been considered
in some previous work. In [
As target group we consider healthy retirees as well as people who are coming closer to their retirement, both having basic familiarity with ICT technologies e.g. Internet, Web browsing, use of smartphones. It is worth pointing out that such target group is not homogeneous. Indeed, some of them may have limitations in using technology depending on e.g. cognitive issues (memory, language comprehension, visual attention, ..), perception problems (vision, audition) and motion problems, not mentioning that such abilities can also decline over time. In addition, target seniors span an array of different age, background and knowledge, and may be even considered differently depending on the sector in which they work/have worked. For example, in the information technology sector, a main concern for a potential employer is not represented by user motor impairments, but rather by his/her cognitive abilities and knowledge skills. Therefore, suitable adaptation /personalisation techniques should be put in place to address such highly varied yet specific users’ characteristics.
From a technological point of view we have designed a platform including a number of modules described in the continuation of the paper. In the following section, an example scenario of use is provided for the expected system.
An Example Scenario Giovanni is a 70 years old man, former employee of an ICT company. He retired two years ago. Three months ago his wife died: since then, he is facing an increase in his monthly expenditure (e.g. he pays someone to fulfil the home needs previously done by his wife). He was told by one friend about the existence of a platform through which companies can indicate consultancy/knowledge transfer needs they might have and which could be filled by an older candidate. Giovanni decides registering to the platform and fills in some information about himself. After some days, the platform informs him of an opportunity posted by a small company, which cannot afford the high costs of a specialized assistance. Giovanni logs in the system from his PC and receives further details on the offer through an adapted, multimodal UI. Giovanni is a bit sceptical about his capability to cope with that challenge: the system detects a lack of reaction and adapts its behaviour accordingly, by providing him with further info about the advantages of accepting the offer, e.g. possibility to work from home, with flexible time arrangements and, if selected, he will be paid 2000 euros, a good complement to his pension. Giovanni finally accepts the offer and starts drafting a solution. The platform, according to the information associated with him (e.g. knowledge, preferences, skills and abilities) adapts its UI so that he can easily submit his proposal in the system. Three weeks later, Giovanni is informed that his idea was selected. Apart from being satisfied by his personal achievement and regaining self-confidence, he also thinks that the monetary reward are very welcome. Finally, the monitoring system of the platform, which tracks Giovanni’s work/social activity and health status, records his improvements in the last period and appropriately informs him about them. It also rewards and further encourages him to stay active, by highlighting the overall benefits of this good behaviour.
THE PLATFORM ARCHITECTURE The platform aims to support a variety of services for active elderly, e.g. sharing experiences and content across their communities, connecting job seekers that offer jobs for elderly, as well as persuading users to stay active through adequately adapted Web interfaces supporting anytimeanywhere access to the platform services.
The architecture of the platform is mainly based on Webbased technologies to allow the elderly to access and use anytime and anywhere the platform from various devices (PCs, tablets, smartphones, TVs, ...). It consists of a number of modules providing the needed functionalities, which are split into User-oriented services and Core Services in Figure 1. The latter is the set of logical modules providing basic functionalities mostly exploited by other modules of the platform. The User-oriented services have instead a more user-related dimension, i.e. being either platform services directly exploited by users, or services exploited by the system applications.
Manager can occur e.g. over HTTP (i.e. in case of a JavaScript Context Delegate, an Ajax call can be used). The Core Services mainly include Adaptation, Behaviour Analysis and Context Manager. Context information is gathered by the Context Manager which has a client-server architecture. The context manager server receives the contextual information from the Context Delegates associated with each user/device, and which are able to supply the context data detected by the sensors connected to the devices. A Context Delegate can be implemented in different ways according to e.g. device capabilities. For instance, the Context Delegate for monitoring user position could be a lightweight stand-alone application that reads the GPS serial stream, extracts the terrestrial coordinates and forwards them to the Context Manager, while the Context Delegate for tracking user’s Web activity can directly be included in the navigated Web page as JavaScript code. The connection between a Context Delegate and the Context The Context Manager server uses an associated repository (Context data) for e.g. storing and updating four types of context information: user (preferences, tasks, physical and emotional state, …), devices (interaction resources, connectivity, …), environment (light, noise, ..), and information concerning social relations (groups of users, privacy rules, …).
All the data collected is gathered by the Context Manager
server and stored in the Context Data repository, which is a
collection of context entity instances modelled according to
a specific XML-based language. The Context Manager
provides an interface for performing typical operations on
such repository (e.g. insert/query, update/remove a context
entity, (un)subscribe to receive updates on changes of
context entities’ state).
The Context Manager also provides the Behaviour Analysis
module with relevant context information, so that the latter
module can assess whether any abnormal situation has
occurred in users’ behaviour. The Behaviour Analysis
module gets information about users (User models
repository), their planned activities (Task models
repository, represented using the ConcurTaskTrees notation
[
The User-oriented services include two main functionalities and two main data repositories. The data repositories are the repository for handling privacy rules and the repository containing structured information on social knowledge. Both of them will be used by the Social Features module, which includes social-oriented services, such as features supporting users to share/annotate relevant content based on their experiences, crowdsourcing services providing relevant functionalities for handling task/job offers from organisations/job providers and then connect such offers with skills, knowledge and abilities of elderly through an adequate match-making strategy. It is worth noting that users will be able to exploit not only the applications (built on top of platform functionalities), but also directly some services of the platform (namely: crowdsourcing service, social knowledge sharing and persuasion service). A number of applications can be identified (e.g. sharing personal knowledge on multimedia TV content, supporting workers having initial health diseases). They will be developed on top of the services provided by the platform. Monitoring and Behaviour Analysis The literature reports that electrophysiological body signals (e.g. galvanic skin response or skin conductivity, heart rate, facial electromyography, electroencephalography, blood pressure and respiration) are related to personal emotional state. Thus, one of the goals of this module will be to monitor the elderly’s status so as to be able to develop algorithms for analysing/handling their emotional state. Moreover, this module will perform an analysis of users’ behaviour in order to detect whether abnormal deviations occur. The analysis is carried out by comparing elderly routine/planned/expected user tasks, with the behaviour that they actually show. Routine tasks are specified in task description repositories (namely: CTT task models, which are specified in XML, a format which supports high flexibility/interoperability), while the actual behaviour of the elderly will be captured through various types of ubiquitous/body sensors. The analysis will be mainly carried out by comparing the expected actions (specified according to the hierarchical CTT task model) and the sequence of logged actions corresponding to the actual user behaviour, which are gathered in the current context and provided by the Context Manager. The outcome of such a comparison will be able to highlight a range of possibilities about where the main problem is, for instance performing actions in a wrong order, not performing at all an action, performing an action too many times, performing the action using incorrect resources, and so on. The various outcomes will then be classified in terms of importance/seriousness and adequate actions will be undertook accordingly by the platform (e.g. sending some messages to the user). Identifying elderly’s decline in cognitive/physical/social activity Functional decline and chronic illness become increasingly prevalent/likely with advancing age. Our idea is that, by comparing the usual elderly’s routines with tasks they actually do, or more in general, by analysing their health status, it is possible to spot in advance early symptoms of a mild/progressive decline in their physical, mental/cognitive and social status. This, in the long term, could compromise their willingness/ability to remain active, then it is important to identify such deviations early so as to be able to act promptly.
Mental/cognitive/emotional changes. A typical example of initial mental/cognitive decline can be represented by a non-adherence with medication, e.g. elderly start to mismanage their daily medicines (alter the doses, swap medicines), or they even forget to take them. If this occurs frequently it could be a sign of a mental decline. Another example considers the fact that older people often link the administration of such medication to specific lifestyle events, time and patterns of daily activities: involuntarily and frequently altering these patterns can be a symptom that the elderly is mildly/progressively experiencing a mental decline. Other deviations can occur when the elderly e.g. place objects in the wrong place, or when they are unable anymore to carry out common/essential daily activities autonomously, as they used to do before. Other changes could involve emotions: feelings of sadness over an extended period of time could indicate depression. However, it is worth pointing out that depression could also be connected with e.g. elderly not eating regularly anymore and/or staying in bed longer than usual.
Physical changes. An abnormal behaviour could be identified in a progressive reduction (or even a lack) of physical activity, or a deterioration of elderly physical/muscular strength (e.g. the elderly is less able to grasp common objects with a hand, or his/her moving is becoming slower than before). Another example of a physical decline can occur e.g. when the elderly experiences more frequent awakenings during the night (e.g. to go to bathroom). This could lead to have a poor sleep, which in turn could conduct to memory declining and brain deterioration.
Social activity changes. Elderly decline in social standing
include e.g. having a smaller social network, less frequent
participation in social activities, less interaction with
friends, etc. Social disengagement/isolation/exclusion could
have consequences also on elderly’s mental decline.
Above, we just provided examples of abnormal situations
that could be signs of a starting elderly’s decline. They
could be identified by comparing suitable descriptions of
expected elderly activities (e.g. contained in CTT task
models) and/or status, with actual elderly
activities/conditions. Of course, the situations actually
detected will depend on the specific type of sensors used.
The Context Manager
The sensed context data will be made available to other
architectural components by the Context Manager, which
consists of a set of (REST) Web Services providing access
to various operations (e.g.: subscribe, update, notify) for
handling context data. Thus, it will act as a centralised
mediator between entities providing context information
(e.g. sensors), and entities consuming context information
(e.g. services). The Context Manager will also store and
access in a consistent and flexible manner rough context
data acquired from heterogeneous and arbitrary
sensors/devices/applications in the user’s environment, and
then aggregate/abstract these data into relevant context
information, so as to make it available to other
contextsensitive components of the platform architecture.
The Adaptation Module
This module supports relevant context-aware techniques
able to support runtime adaptation depending on the
characteristics of e.g. users (characteristics, limitations,
knowledge, skills, emotional state), available devices,
surrounding environment. In order to be able to express
various types of adaptations, a specific language for
adaptation has been developed [
Social Features This module provides intuitive social/knowledge sharing mechanisms empowering seniors to e.g. share their experience/expertise with others, facilitate information transfer/communication, and get in contact with other people. This will have the benefit of e.g. helping the elderly in creating and maintaining social relationships and contacts, while limiting the risks of their social exclusion and isolation. In addition, the social/knowledge sharing features will support a process in which the elderly will be engaged in informal learning activities where people will be able to learn from life experiences of seniors, even in intergenerational interaction. Moreover, this module will provide some Web-based crowdsourcing services supporting elderly in having an easier and more effective access to job/activity opportunities specifically targeted to them will be developed. Through this crowdsourcing services potential job providers (e.g. public/private organisations, industrial companies, etc.) can specify tasks and/or particular expertise/knowledge/skills they need. According to requested skills, users of the virtual elderly crowd will be able to respond, by offering their availability for filling that gap through their contribution/work. The crowdsourcing services will be built in such a way to support match-making between skills required and competences offered, as well as facilitate even nontechnologically skilled senior people to access, be part and actively engage in the virtual elderly crowd.
Persuasion Services This module supports effective motivational strategies to stimulate seniors’ behaviour change, by using just-in-time, context-dependent information, and present at the appropriate time/place simple and tailored triggering messages, to persuade elderly to have a target behaviour. Persuasive techniques will also be realised through positive reinforcement/rewarding strategies delivered to seniors when they perform the desired behaviour. They can be realised by using e.g. virtual points/scores or monetary incentives and/or by reminding users how good it is to them a certain behaviour and/or by providing them with suitable visualisation techniques for supporting selfawareness/monitoring of their current behaviour progress. CONCLUSIONS In this paper we have presented the architecture of a Webbased platform for stimulating the elderly to stay active in their life. The platform consists of a set of logical modules supporting core and user-oriented services.
Future work will be dedicated to developing some applications for active elderly able to exploit the features of such platform and to provide opportunities for end user validation.
ACKNOWLEDGMENTS We gratefully thank the SERENOA ICT EU Project for the partial support in the development of some components of this architecture.