SVARNE, an Expert System
To Be Used in the Care of The Mentally Retarded

Charlotte Magnusson, Jan Eric Larsson, Arne Svensk

Proceedings of the 3rd European Conference
on the Advancement of Rehabilitation Technology (ECART 3),
pp 198-200, Lisbon, Portugal, October 1995




 

Abstract

Designed as a system which, by transfer of knowledge and extended documentation, can act as one of many tools when it comes to dealing with situations containing violence, the SVARNE system may turn out to be an useful aid not only for people working within the care of the mentally retarded, but most importantly for the handicapped persons themselves. This paper presents the SVARNE system as it looks today and indicates some future developments.

Introduction

People with cognitive dysfunctions combined with a limited ability to communicate are almost completely at the mercy of their environment (parents, relatives, staff, etc) when it comes to recording their personal history. While most people have access to devices which they can operate themselves and which produces written accounts, pictures or videotapes that can be used to remember (and tell somebody else) what happened, this group of people often has to rely on other people to help them with this.

It has recently been pointed out [2], [3], [4] that this documentation, particularly when it comes to the transfer of knowledge about the life history and social network of a person, in most cases is fragmentary (even important facts/events may be completely left out).

Knowledge of the life histories and social networks improves the possibilities for communication as well as the general understanding of a person and his or her behavior. Having access to your own life history provides you with increased possibilities of showing who you are and to assert yourself as a person [4]. Thus any system/device/method that improves the documentation of this knowledge is an important aid not only for the people surrounding the differently abled person but also for the handicapped person himself/herself.

The above discussed knowledge becomes particularly important when it comes to dealing with situations containing violence [5], [6], [7]. Violence may actually be triggered by lack of knowledge on the part of the environment, and even though this is not the case a thorough understanding of a person greatly improves the ability to deal with any type of situation containing violent behavior (in this context it is important to point out that the fraction of violent persons in fact is smaller in a mentally retarded population than in a so called normal population).

Because violence leads to such serious consequences, both for the mentally retarded person and the people surrounding him or her, we felt that when it came to deciding on a limited field of interest, where we could test if a computer based decision aid may be one way of documenting and transferring knowledge within the care of the mentally retarded, situations containing violence would have to be our choice.

SVARNE

The notion that expert systems [8] might improve the way knowledge is handled, transferred or stored within the care of the mentally retarded led us to attempt the construction of the expert system SVARNE. The name SVARNE is a combination of the name of the first expert used in the system (Arne) and the Swedish word for answer (svar).

The knowledge implemented in SVARNE is knowledge of structures and relations. When it comes to detailed knowledge or difficult judgements SVARNE relies on the user. The relation between SVARNE and the user is not that of an expert and somebody who knows nothing - instead it is the combined knowledge of the user and the system that makes the expert. It is also important to stress that the knowledge in SVARNE is that of experienced staff, which means that the knowledge comes from within the user-group and not from some expert outside this group.

An important part of the work has been to find a structure in the practical knowledge possessed by experienced staff. This work has turned out to be important in itself as more structured ways of thinking improves the quality of communication and makes it easier to transfer knowledge. The expert system is crucial in the work as it requires exact statements. Because of this it is no longer possible to get away with the vague or "soft" statements that traditionally has been used within the care of the mentally retarded. The use of SVARNE will thus hopefully lead to a better understanding of the problems, a more reliable and clearly expressed knowledge, and useful documentation - i.e. knowledge refinement. In this process the expert system itself becomes a tool for extracting structured information.

The knowledge so far implemented in SVARNE concerns violence, but the experience gained in the present work will be useful also when it comes to handling more general knowledge concerning the mentally retarded. Because the nature of the problem (violence) SVARNE has to be run either to prevent outbreaks of violence or after a violent situation has occurred. It is not realistic to expect anybody to run SVARNE and handle violence at the same time. Running SVARNE preventively is useful primarily for acquiring knowledge of how to avoid violent behavior. To run SVARNE after the fact, so to speak, will help the user to find possible reasons for the aggression and thus to avoid future outbreaks of the same kind. SVARNE will also be useful in educational situations. SVARNE has furthermore turned out to be useful when run by two people, one experienced and one beginner. In this case the experienced person contributes with the deep knowledge and detailed stories while the computer provides the systematic structure behind the conversation/exchange of knowledge.

The intention behind the present project is that the discussions around SVARNE will help the structuring of hitherto unstructured knowledge within the care of the mentally retarded. Running SVARNE will provide a common structure which will be useful in all kinds of communication concerning violence and thus, hopefully, improve the care given to the persons in question. The reminder or "thought support" provided by the questions in SVARNE is the important thing even though the actual advice given by SVARNE in many cases will prove just as useful. The work is distributed between the user and the computer in such a way as to let each do what they do best, respectively: the computer reminds about details and enforces a systematic check, while the human is involved in and in charge of all important judgements and decisions. In this way the human and the computer will cooperate in a good way and together give a better result.

Another important aspect of SVARNE is that the work will give good documentation both of the general knowledge and of the different cases. The system itself provides possibilities for documentation, as it contains a personal data base as well as a structured way of entering rules specific for one person.

SVARNE exists today at six group homes in the south of Sweden. The people working at these group homes have been the experts whose knowledge we have attempted to implement in the program. SVARNE has also been placed for evaluation at a seventh group home, which previously has had no contact with the project.

Conclusion

SVARNE has shown the expert system approach to be useful also when it comes to handle as complex knowledge as that associated with problems occurring within the care of the mentally retarded. The good results obtained in the project also indicate that the procedure used to extract knowledge (i.e. working iteratively in a group where the roles of experts and knowledge engineers overlap) is particularly useful when approaching areas of "soft", unstructured and/or complex knowledge.

A further result is that the check-list type of expert system, where the user has to take active part in the decisions, is particularly useful for the present kind of knowledge. The gain is both practical and psychological. Practical because some difficult decisions are almost impossible to implement in an expert system and psychological due to the fact that the machine is not telling you exactly what to do (and reducing the user to a rather unimportant figure who just carries out the wishes of a computer) but rather acts as a support to the decision making of the user (where it is the user that is, so to say, in control). In the care of the mentally retarded one has traditionally been rather uninterested in, and sometimes even hostile to, using technological devices. It is thus an important indication that the design used in SVARNE is well suited to the problem when, of all the people who have been shown SVARNE, not one has voiced any serious criticism but instead expressed a very positive attitude to the project.

The long range goal of SVARNE is to improve the way knowledge is handled and stored within the care of the mentally retarded and in this way also to improve the care given to this group.

Future developments

The evaluation of SVARNE has highlighted the need for a network between different group homes. One part of the work with creating a more general network is to set up a network of computers. Such a network is at present under construction, and will be of great use not only for employees but also for the mentally retarded people living at the different group homes.

Another development is to adjust SVARNE in the direction of creating history books. When running the SVARNE system one would in this case get history book information, case book information, intellectual ability information and finally the decision support (where only questions relevant considering the history book, case book and intellectual ability information should appear).

As the history book information will be extensive, it is probably not practical to display all of it when running the SVARNE decision support - instead particularly important events can be displayed, while the rest of the history book is accessible via a special editor/tool.

One further development is to investigate the usefulness of SVARNE within the field of psychiatry. This work has recently been initiated, and it is hoped that it will lead to an improved general decision support concerning violence.

Acknowledgments

The authors gratefully acknowledge useful suggestions from Eric Astor, Ingrid Liljeroth och Bodil Jönsson.

References

[1] Isaac - A Personal Digital Assistant for the Differently Abled, B. Jönsson, A. Svensk, Proceedings of the 2nd TIDE Congress, 1995

[2] Expertsystemet SVARNE - beslutsstöd vid våldssituationer, C. Magnusson, A. Svensk, CERTEC/LTH 4:93, 1993

[3] En undersökning om hur den enskilde omsorgstagarens historia tas tillvara i ett gruppboende, M. Carlsson, M. Gunnarsson, report from Kommunala högskolan, Malmö, Sweden, 1993

[4] Albumet Min Historie - ett nytänkande från Norge, article in Nordisk Utveckling 4:1994, about the album "Min Historie", K. Gundersen, T.S. Olsen, L. Moynahan, S. Solberg, Haugtussa forlag, Norway

[5] MOSAIK: Om Projekt "Vold som udtryksform", Socialt Udviklingscenter, Storkobenhamn, Randersgade 17, 2100, Kobenhavn O, Denmark

[6] Vägval och växande - specialpedagogiskt perspektiv, Danielsson, L., Liljeroth, I., Utbildningsförlaget, 1987

[7] Att förstå och påverka beteendeproblem, Wadström, O., Aros förlag, Danmarksby, 755 98 Uppsala, 1989

[8] Encyclopedia of Artificial Intelligence, vol 1, Shapiro, S. C., John Wiley & Sons, Inc., 1992