The Carer App is a mobile creativity support app with which to care for people with dementia. Care staff can use it to create new solutions to challenging behaviours through case-based and analogical reasoning with recorded good care practices. To enable this to happen the app implements computational creativity services in order to generate content to support cognitive creative tasks by care staff.

Target users: RNHA - Care staff working with residents with dementia.

Contact for the Carer app:  This e-mail address is being protected from spambots. You need JavaScript enabled to view it

Link to the video

This video describes the Carer App

Link to the test version

If you would like to have a look at the app there is a Carer app testversion.

How does it work?

More contextualized support for care staff to aid them with the diagnosis and response to difficult situations. When faced with a difficult situation care staff will need advice and information quickly, delivered in situ to the care to be given, and available around the clock. Some emerging good practices are available. For example the Social Care Institute for Excellence report prototypical examples of good care practice in the form of case studies composed of the problem situation encountered and the observed good practice that resolved the situation [2]. Each case study is described in 200-300 words. In MIRROR we conjecture that these short case studies can provide a structure and format with which to capture then share contextualized good practices to care staff in situ. Therefore we are delivering new apps and services to run on mobile devices that will allow care staff to describe a problematic situation that they are encountering in natural language, retrieve similar case studies automatically in a matter of seconds, read then browse the case studies to select one or more appropriate to the situation, then implement good practice based on the recommendations provided. To be able to deliver these apps and services we are exploiting sophisticated natural language parsing and matching technologies developed originally for software design but well-suited to this challenge.

Carer’s Discovery Algorithm
The repository contains structured descriptions of case studies in an XML data structure based on the structure of case studies reported by the Social Care Institute for Excellence [2]. Each has two main elements, both expressed in natural language – the problem encountered and solution applied.
The algorithm discovers and retrieves descriptions of case studies from problem queries expressed in natural language by a member of care staff. Queries expressed in natural language, although easy to write, can be ambiguous and incomplete, so the algorithm has two important capabilities described at length in [3]: 1. Query expansion – the addition of terms in the query that have the same or similar meaning to existing query terms, to make the query more complete; 2. Term disambiguation – selecting the meaning, or sense of each term in the query to enable query expansion, thus making the query unambiguous.

For example the query Mrs. X acted aggressively towards other residents at breakfast. Suspect underlying insecurities to new people is incomplete because it does not define the aggressive behavior observed, and it is ambiguous because it does not define what the meaning of underlying insecurities. There are several possible meanings of insecurity, for example the state of being subject to danger or injury, or the anxiety you experience when you feel vulnerable and insecure? To make this query more complete expansion techniques are applied to generate more complete case queries, and term disambiguation techniques from information retrieval are applied to generate unambiguous queries. The algorithm has the four key components, and the WordNet on-line lexicon [4] fulfils an important role for three of them. In the first the query is divided into sentences then tokenized and part-of-speech tagged and modified to include to resident). In the second the algorithm applies procedures to disambiguate each term by defining its correct sense and tagging it with that sense by iteratively using context knowledge from other terms in the query (e.g. defining a resident to be a someone who lives at a particular place for a prolonged period rather than a who lives in a hospital and cares for hospitalized patients under the supervision of the medical staff of the hospital). In the third the algorithm expands each term with other terms that have similar meaning according to the tagged sense, to increase the likelihood of a match with a case study (e.g. the term aggressive is synonymous with the term hostile which is also then included in the query). In the fourth the algorithm matches all expanded and sense-tagged query terms to a similar set of terms that describe each case study in the case study repository. Query matching is in 2 steps: (i) XQuery text-searching functions to discover an initial set of case study problem descriptions that satisfy global search constraints; (ii) traditional vector-space model information retrieval, enhanced with WordNet, to further refine and assess the quality of the candidate case study set. The output is a ranked list of matched case studies in more-relevant to less-relevant order.

The Interactive Carer App
The Carer app that care staff use to generate problem queries and browse retrieved case studies was developed using ASP.NET and the open source iWebKit framework for the iPod touch [5]. A user enters a natural language description of a problem situation using the text box shown on the left hand side of Figure 3, then refine it by selecting one of the observed behavior values such as Resident is violent shown on the right. When the user presses the Retrieve similar cases button the app automatically generates a query that it fires at the case study repository.
More Details about the app can be found in Karlsen et al. [1].

References
 [1] Karlsen, K.; Zachos, K.; Maiden, N.; Jones, S.; Turner, I.; Rose, M.; Pudney, K.; , "Supporting reflection and creative thinking by carers of older people with dementia," Pervasive Computing Technologies for Healthcare (PervasiveHealth), 2011 5th International Conference on , vol., no., pp.526-529, 23-26 May 2011
[2] Owen T. & Meyer J., 2009, ‘Minimising the use of ‘restraint’ in care homes: Challenges, dilemmas and positive approaches’, Adult Services Report 25, Social Care Institute of Excellence
[3] Zachos K., Maiden N.A.M., Zhu X. & Jones S., 2007, ‘Discovering Web Services To Specify More Complete System Requirements’, Proceedings CaiSE’2007, Springer-Verlag Lecture Notes on Computer Science LNCS 4495, 142-157
[4] Miller K., 1993, ‘Introduction to WordNet: an On-line Lexical Database’ Distributed with WordNet software
[5] iWebkit 2011