Show Summary Details
Page of

The use of smart technology in the management and rehabilitation of executive disorders 

The use of smart technology in the management and rehabilitation of executive disorders
Chapter:
The use of smart technology in the management and rehabilitation of executive disorders
Author(s):

Roger Orpwood

DOI:
10.1093/med/9780198568056.003.0013
Page of

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). © Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy).

Subscriber: null; date: 12 December 2017

Introduction

There has been a growing interest in the use of intelligent homes to support people with a variety of disabilities (Bjørneby 1997; Dewsbury et al. 2004; Orpwood 2006). Initial work has focused on supporting people with physical disabilities (Gann et al. 1999), and these studies clearly showed the potential of such approaches. Most work on intelligent homes has used the established devices that have been available for automatic environmental control for many years and applied them to supporting the new client group (Porteus and Brownsell 2000). For people with physical disabilities, this approach was sensible and provided many benefits. However, when attempts were made to apply the same approach to people with cognitive problems it was clear that some rethinking was necessary. For example, one very useful development concerned the operation of bathroom taps. Taps that can be linked to smart home environments and can be operated purely by moving your hands under the tap have been available for some time. An infrared beam is activated that initiates the flow of water. Such remote activation is really useful for someone with poor hand function from, say, arthritis. However, such operations can be very confusing for someone with an executive disorder leading to cognitive problems. Understanding that moving your hands under the tap is needed in order to make water come out is cognitively challenging, and not part of common experience.

Therefore a new approach is needed for people with cognitive problems. The work to support people with cognitive problems that is being explored at the Bath Institute of Medical Engineering (BIME) has taken a very needs-led approach. Most of the original work has been carried out with people with dementia, but the basic principles apply to anyone with executive control problems. Rather than simply matching people against the technology that was available, the approach was to explore the needs of the client group in the first instance to obtain an understanding of the key issues that needed to be addressed (Jepson and Orpwood 2000). This understanding was then used to define potentially useful technology. If established technology were available, clearly this could be used, but there were many situations where it was quite obvious there was nothing suitable available, and a programme of new design and development was initiated (Orpwood et al. 2005). Therefore the work was very much needs led rather than technology led.

In this chapter we explore the way that this technology has been developed to support people with cognitive problems, and in particular executive disorders. We examine the kind of technology that is needed, and look at the results of some initial work in applying it. We discuss how this potentially very beneficial technology can be tailored to suit the needs of this client group.

What is smart home technology?

The smart home descriptor is somewhat over-employed these days, with many technologies using the ‘smart’ label to imply something with a bit of intelligence. Of course, manufacturers have jumped on the bandwagon in the hope that the label imbues some added value to their products. This has happened, for example, with Telecare technology where the manufacturers have been using the ‘smart house’ label for the last few years, even though Telecare is far from having much intelligence. It is often no more than a basic community alarm system that involves a round-the-clock call centre able to respond in an emergency or provide regular contact by telephone. The work described in this chapter uses technology that does include a level of intelligence, and is able to provide autonomous assistance rather than simply detecting problems and calling for help via call centres or through nurse call systems.

The infrastructure for providing homes with autonomous capabilities requires three key components: behaviour monitoring sensors, assistive support technology, and a communication link between the two (Figure 13.1).

Fig. 13.1 The components required for an autonomous smart home. Outside help is only required in the event of the home not being able to deal with problems itself.

Fig. 13.1
The components required for an autonomous smart home. Outside help is only required in the event of the home not being able to deal with problems itself.

Sensors

Sensing technology is used to monitor the behaviour of people in the home, or the way they interact with their environment. Many sensors are already available for the mainstream application of the technology. Components such as movement sensors using passive infra-red (PIR) units, similar to those used for burglar alarms, are well developed. For similar applications, sensors are available to monitor whether windows or doors are open. Further sensors have been developed as part of the current interest in Telecare technology, such as bed-occupancy sensors. However, there is an important need for further work to develop sensors that are specific to the needs of this client group. Some, such as toilet-usage monitors and fall detectors, are the subject of much current research (e.g. McKenna and Nait Charif 2005).

Support devices

Support technology is used to react to the information being provided by the sensors, and to try and assist the occupant to maintain independence and a good quality of life. It is in the area of support devices that most work is now needed. Again, some devices, such as automatic cooker shut-off valves (Gibbs et al. 2003), are fairly straightforward, but others, such as bath tap shut-off devices that do not take control away from the user, are still quite experimental. An important area of work, described in more detail below, is a means for providing prompts and reminders to occupants.

Communication bus

A key component of all autonomous installations is a means for all the sensors and the support devices to talk to each other. This communication is provided through what is known as a communication bus. This bus can simply be a set of extra wiring that links all the components together, but it can also be via a radio and, therefore, wireless link. The bus does rather more than just allow a sensor such as a bed-occupancy sensor to turn on a bedroom light. It has computing facilities embedded within it to enable judgements to be made about appropriate actions on the part of the support devices, depending on what the sensors are telling it. It also provides a means for coded messages to be sent from the sensors to the whole smart installation so that any component can know what has just happened. It also has facilities for checking that the message it has just sent out has been received, and will send the message repeatedly until this is assured. These systems were developed for providing automatic control in large public buildings such as airports, and needed to be extremely reliable and effective, and require minimal intervention. Therefore they are ideal for the kinds of application described here where someone’s safety or security is at stake. The work undertaken in Bath has used a communication bus known as the European Installation Bus (EIB), which is now incorporated in the KNX system. It is the most popular system in Europe and many components are available, such as different sensors that can work with it. However, radio-based versions have only been made available recently, and the range is still rather limited. Other systems are in use, and one still evolving which looks to have much promise for these kinds of applications, particularly where radio communication is involved, is a system known as Ziggbee.

To recap, the sensors monitor the user’s behaviour, and this monitoring provides the basis for judgements to be made about how the user can be helped. The communication bus is able to listen to all the sensors and to talk to appropriate support devices to initiate their action, depending on the judgements made by its embedded computer. In many ways the effective function of an autonomous smart installation is to emulate the behaviour of a good carer, but of course without getting tired or frustrated or emotionally involved in other ways. However, what the technology cannot do is to replace the love, affection, personal understanding, and social interaction that a carer can provide. Some have criticized the work for that reason. If service providers see the technology as a complete replacement of human carers, possibly with a view to saving money, then this criticism would be valid. However, most can see that there is a lot of potential for this kind of technology to augment personal human care, and in many ways it can be more effective in situations where constant close monitoring is needed. It can also provide the client with some dignity and self respect by giving them some independence and low-level back-up for daily activities, without having to rely on human carers or interventions from others to resolve their difficulties.

What kind of support can be provided?

Examples of technology

What sort of support can this technology provide? A few examples can illustrate its potential. Sensors can detect user’s movements around their living space and use this information to provide support. This support could simply be turning off appliances that had been forgotten about, like the cooker or the bath taps. It could provide automatic lighting when someone gets out of bed at night to provide orientation and help prevent falls. However, such systems really come into their own when they are linked to voice prompting devices. For example, if someone was confused about time and tended to go out of the house at inappropriate times, it is quite easy to detect their movement near an external door and, knowing that it is an inappropriate time to go out, to prompt them with a message to that effect. Similarly, if someone was prone to getting up in the middle of the night and becoming a little disoriented, their movements and the fact that they were out of bed could be used to provide them with a prompt about the time and to suggest that they go back to bed. If a user had a habit of putting something on the cooker and wandering off, they could be prompted when they left the room. The house could intervene to turn the cooker off if smoke were detected or the kettle had boiled dry, or with appropriate sensors turn off a gas cooker if the gas had not been lit.

Voice prompts

When voice prompts were originally planned some concern was expressed about how people would react to them. It was felt that users with cognitive problems would become anxious and a little alarmed about voices coming out of nowhere. When the use of such prompts was originally explored, the voices came out of appliances that people expected voices to emanate from, such as the TV or the radio. This was quite difficult to configure but could be done. If the radio was off it would be automatically turned on to play a message. If it were on, and playing a broadcast, it would be overridden to play the message. However. some work was carried out using simple speaker boxes, and these were found to be just as effective (Orpwood et al. 2004). Users were not at all concerned about the disembodied voice. They could see that the voice was coming out of a box, and it was part of their common experience that voices come out of things like tape recorders and record players.

It was unclear as to what would be the best voice to use. Should it be a known voice or an anonymous one? Should it perhaps be the person’s own voice? A concern was expressed that the voice ought not be recognizable as the user might think the speaker was in the home, and go looking for him/her.

For this reason, all the early voice prompts used a warm anonymous voice. However, it was found that people do not behave inappropriately if they hear a voice that they recognize. On the contrary, it appears that people respond better to a voice they trust, which inevitably means that it is recognized. Recent work with a client with quite severe dementia (MMSE score of 10) has found that he has no problem with his daughter’s voice coming out of speaker boxes mounted in the rooms (Evans et al. 2007). He knows that she is not really there, but he also knows that she cares for him and would give him good advice. When the system was set up various messages were recorded for him to use. His daughter was asked to imagine she was standing next to her father when, for example, he was about to go outside in the middle of the night, and to talk to him in a way that she felt would prompt him to respond. Her messages were quite assertive and even a little angry, and there was a concern that they might just upset or alienate him. However, they have worked very well, and when asked about her tone of voice the daughter said that she knew that this was the kind of tone that her father would respond to if she had to advise him. Her father is quite accepting and pleased with the messages, and accepts that they help him. The messages would appear to have some emotional value because, despite his dementia, he can relate the content of the messages during subsequent interviews.

Some of this experience of using voice prompts has been applied to people with acquired brain injuries, and the indications are that they are well accepted and acted upon. They have much potential for providing prompts and reminders for people with memory problems.

Flexibility

One of the benefits of the kind of technology that has been discussed is the flexible way in which it can be applied. Of course, people are very different and their needs vary enormously. Their needs will also change with time. Such variability could be seen as a real problem, but the nature of autonomous smart environments is that it is fairly easy to adapt the technology to match the needs of the user. Some users may only need some lighting support at night. Others might benefit from simple prompts for things they tend to forget. Others might require something more substantial to deal with their problems of wandering or forgetful use of household appliances. All can benefit from the same installation, but it has to be configured by simple adjustments to suit their needs.

There are two issues which need to be dealt with to realize the inherent flexibility of these systems. It is fairly easy for an engineer to make the software adjustments once they have been identified, but it would be very confusing for a care professional without engineering skills. There is a real need for autonomous installations to be designed to allow easy changes to be made by intelligent but non-technical carers. The second important issue concerns the definition of the user’s needs. Many aspects will be understood from prior assessments of the user carried out by occupational therapists and psychologists, but there are often issues which only come to light following the detailed monitoring that smart environments can provide. There is a need for careful procedures to be put in place to allow the tailoring of the environment for the individual to be an ongoing process, and for the system to provide care staff with constant reports of how the user is getting on.

Installations

To provide a home with the abilities described requires installation of the sensors, the support devices, and the communication bus. Of these, the one that can cause the most problems and most disruption is the communication linkage. If a new-build care home is being planned, it could make sense to install a hard-wired communication bus throughout new development, terminating the cabling in strategic positions within the rooms so that sensors and support devices can be connected at a later date according to the client’s needs. However, it is extremely difficult to anticipate accurately where such wiring should terminate, and there always appears to be a need for a lot of extra wiring when it comes to subsequently installing the smart equipment. When such hard-wired systems are to be installed in people’s homes the situation is far worse. The installer will probably have to bury wires in the plaster or take up floor boards, and the whole process would be extremely disruptive. This could be a very traumatic experience for clients with cognitive problems.

Radio-based systems have many advantages because of ease of installation and adaptability. Sensors and support devices can be placed where required for a particular client, and therefore retrospective installations in people’s own homes are much less disruptive and far quicker. The downside for radio-based systems is that they require electrical power from somewhere, and this usually means battery power for most of the equipment. The installed devices will indicate when their batteries are getting flat, but the use of batteries means that a battery-checking and replacement policy is required to ensure that they are replaced before they go completely flat and lose their function, with potentially dangerous consequences. Fortunately, battery life is quite good with modern devices, and with the advent of communication protocols such as Ziggbee with excellent power management, the battery life for many appliances would be over a year. The other downside with radio-based systems is the existence of radio dead-spots where communication to a sensor or support device would not be possible. However, a good installer would check such things, and find alternative placements.

In recent developments, the power management issue has been addressed by the use of so-called power scavenging. The power levels needed for, say, signalling that a light switch has been turned on are very low. Power scavenging utilizes the user’s own actions to generate some power. For example, the act of turning on a switch will generate some power, and this can be collected and used to augment the power being supplied from the battery. In this way such sensors could run almost indefinitely without any battery changing being needed. Such approaches will only work for very-low-power situations, such as switches or for movement sensors, and would not be appropriate where large amounts of power are needed, such as tap turning-off devices.

In conclusion, it is likely that the use of radio-based system will be the system of choice for many of these kinds of installations, even in the case of new buildings.

Application of the technology to people with acquired brain injuries

The basic work described above was mostly carried out with people with dementia, but the problems tackled were very similar to those experienced by people with ABIs. Some applications of the technology to this group are described below.

Needs surveys

Much technology has been developed that can potentially provide the basic structure for an intelligent smart home environment for people with ABIs. In order to assess its potential, and to highlight any further developments needed, it is important to try to obtain a better understanding of the needs of such clients, and to assess these needs in the light of the kind of interventions that can be provided. An initial study was carried out using paper records to identify when problems arose during the day for a group of people with brain injuries, and when interventions were necessary to provide support. Originally care staff were asked to keep notes of all the interventions they provided over a month-long period. However, it was found that the records were not comprehensive, and often missed crucial data that would be relevant to the provision of automatic support. The exercise was repeated using observations from psychology assistants over a shorter period. A large amount of data was collected, and was categorized according to the issues that arose. Table 13.1 categorizes the data into various difficulties such as sleeping problems, problems with toileting, etc. It shows that there are many issues that cause problems, and it also provides some quantitative information about their relative importance.

Table 13.1 Analysis of data from paper study of the difficulties faced in the home by people with acquired brain injury

No.

Sex

Age group

Calling for help in the night

Getting dressed approp.

Knowing right time to get up

Taking medication correct time

Remembering correct dosage

Understand reason for taking medicine

Using appliances safely

Choosing approp. food to eat

Keeping home clean

Dealing with laundry

Regulating heating in the home

Handling money

Shopping

Forgetting time/date/year etc

Using the telephone

1

M

51-60

1

2

1

2

3

1

2

2

2

2

3

2

2

2

1

2

M

51-60

1

1

1

2

1

1

1

1

2

2

2

1

1

1

1

3

M

51-60

1

1

2

2

2

3

4

1

2

3

3

4

M

41-50

5

M

41-50

2

3

3

4

1

4

3

6

F

41-50

1

1

1

1

2

1

1

2

2

2

2

2

7

M

21-30

1

1

1

1

1

1

2

1

2

1

2

2

1

1

1

8

M

21-30

1

1

1

3

3

1

2

2

2

2

4

2

2

1

1

9

F

<20

10

M

?

3

3

3

3

1

1

1

3

4

3

2

11

?

?

12

F

31-40

2

1

2

3

4

4

3

3

3

2

2

4

3

2

2

13

M

21-30

1

2

2

4

3

4

4

4

4

3

4

4

4

2

14

M

41-50

1

3

3

1

1

2

1

1

1

1

1

1

1

3

4

15

M

31-40

4

4

4

4

4

4

4

4

4

4

4

4

4

2

4

16

M

31-40

1

1

1

1

1

1

1

1

2

1

1

1

1

1

17

M

41-50

1

2

1

2

1

2

3

2

3

2

18

M

51-60

2

2

1

2

2

2

2

2

2

2

2

1

19

M

41-50

1

1

2

2

4

2

2

2

3

3

3

2

2

2

2

20

M

21-30

1

2

3

1

2

1

1

1

4

3

1

21

M

21-30

1

2

2

1

1

1

2

2

1

1

1

3

1

22

F

41-50

1

1

1

2

1

1

2

1

2

2

1

23

M

21-30

3

3

1

3

4

24

F

31-40

1

2

2

4

4

4

3

4

4

4

4

3

3

3

1

1

NOT A PROBLEM

14

9

10

5

5

7

3

10

4

6

5

6

4

4

9

2

OCCASIONAL PROBLEM

3

7

5

4

1

5

8

6

8

6

3

6

8

7

6

3

REGULAR PROBLEM

2

3

5

2

3

0

2

2

2

3

2

3

2

8

2

4

VERY SERIOUS

1

1

1

3

4

4

3

3

3

2

4

4

3

2

3

TOTAL SCORE

30

36

39

31

32

33

37

40

38

35

33

43

38

50

39

NUMBER OF PEOPLE

20

20

21

14

13

16

16

21

17

17

14

19

17

21

20

AVERAGE

1.50

1.80

1.86

2.21

2.46

2.06

2.31

1.90

2.24

2.06

2.36

2.26

2.24

2.38

1.95

Automatic lifestyle monitoring

The data collected from these exercises was useful in that it identified issues to be tackled, but it did not provide very detailed information about their nature or about the user’s behaviour in relation to them. Therefore a further exercise was carried out using automatic data collection via sensors. Such systems provide very good lifestyle-monitoring capabilities, and can show the way that the user is interacting with their environment in great detail. Several sensors have been developed for this work. These include movement monitors based on PIR sensors, bed-occupancy monitors, and monitors to detect door opening. Each sensor contains a small data logger that records all the events that triggered it, and when they occurred. The loggers can subsequently be attached to a computer to download the information stored. By using the data collected from all the sensors a picture can be generated of the user’s behaviour during the monitoring period.

Figure 13.2 shows some typical data for a client in a sheltered environment who was having problems with sleeping. As can be seen he was out of bed frequently during the night. Some of these episodes were clearly to go to the toilet, but during many he just wandered about his apartment. A paper record would just have reported that he had had a disturbed night, but the lifestyle monitoring enabled a much clearer picture to be generated of how he had behaved during the night and gave much more insight into the problem he was experiencing. Such data can be very helpful for making judgements about what sort of interventions would be helpful for clients because it is very detailed, and by definition objective. Of course, it still requires judgements to be made about what kind of intervention might be supportive, but it enables those judgements to be based on a much clearer picture of what is going on. The other major advantage of using such lifestyle monitoring is that, once interventions have been provided, it can provide a great deal of data about how effective the interventions have been, and can be used to quantify their outcome.

Fig. 13.2 (a) Bed occupancy shown for a 24-hour period starting from midnight. (b). Details of bed occupancy and movements within an apartment for a 10-minute period during the night.

Fig. 13.2
(a) Bed occupancy shown for a 24-hour period starting from midnight. (b). Details of bed occupancy and movements within an apartment for a 10-minute period during the night.

Sensors can uncover unknown behaviours

Another client in a sheltered housing scheme had such an installation provided. His assessment prior to moving in showed that he had some continence problems and a tendency to wander at night, but was mostly quite inactive. A standard set of smart technology had been installed in his apartment, and prior to any of the support devices being activated the sensors were used to record his behaviour in a similar manner to that described above. In this case the house had been fitted with a compete smart home infrastructure, and so the computer embedded within in it could directly log all the information provided by the sensors, as well as its own support activities. The initial recording period showed that his behaviour was much more complex than had previously been understood (Evans et al. 2007). For example, he did interact with the cooker, although his family and care-staff were quite adamant that he never did any cooking. His night-time wandering was also quite severe, and his behaviour when he got out of bed also showed signs of confusion, particularly about the location of the toilet. He seemed to follow a very complicated path through the apartment before he went to the toilet (Figure 13.3(a)). These data were invaluable in providing an understanding of his difficulties and for planning appropriate interventions. What came as a complete surprise was that he was getting very little sleep at night (average of 3.53 hours per night). With automatic lighting at night, including toilet guidance and messages from his carer to encourage him to go back to bed rather than wander about the apartment or to go outside, his sleep increased to an average of 5.39 hours per night (Figure 13.3(b)) and his wandering out of the flat markedly reduced. He also had far fewer toileting accidents.

Fig. 13.3 (a). Complex movements around an apartment, and bed occupancy, from midnight to 2 a.m. for a client showing signs of confusion, before support equipment was turned on. (b). Bed occupancy before and after turning on support equipment.

Fig. 13.3
(a). Complex movements around an apartment, and bed occupancy, from midnight to 2 a.m. for a client showing signs of confusion, before support equipment was turned on. (b). Bed occupancy before and after turning on support equipment.

The sensors and the data-logging used for the work described above were purpose designed for the project. However, a number of other groups are exploring this kind of lifestyle monitoring and ways in which the data can be presented to give care-staff a clear view of what has been going on (e.g. Perry et al. 2004), although there are clearly problems in interpreting such complex behavioural data (Hanson et al. 2007). Some of this work has translated into commercial products. A system known as JustChecking has been quite successful. It uses a small series of sensors that can be simply mounted on the wall using self-adhesive Velcro. The sensors have a radio link to a centralized data logger which sends the information to a secure web-site. The data can then be downloaded onto a PC for display. This kind of technology is clearly going to be a very useful tool for care staff to ascertain in more detail how their clients are getting on, and to assess the outcome of interventions. A major issue with this technology is how the data is presented. For research purposes it is important to obtain quite detailed indications of how someone is behaving on a second-by-second basis. However, for use in a care setting it would be more important to provide automatic interpretation of the data that would be meaningful to care staff and the kind of understanding they require. For example, some processing of the data that would indicate that the client was feeling anxious or restless, or flagged up how long the client had been sleeping, would be much more useful. Some work on manipulating such data for these purposes has been completed (Evans et al. 2007), but much more needs to be done in this area in collaboration with care staff. A large amount of useful data can be collected, but it will all go to waste unless some kind of intelligent automatic interpretation is possible.

The need for close involvement of users

It would be useful to relate a case study that illustrates very well the importance of close user involvement to enable appropriate support to be provided. A client reported being restless at night, and staff complained that he would often wander out of his room. A wander reminder was tried with him. These devices just detect movement near a door during the night and replay a message to discourage the client from going out. He was happy to have one installed but it did not really make much difference. He would still go looking for staff in the night. Discussions with him indicated that his sleeping was severely affected by night-time anxieties. He reported that he would often wake up with some deep concern that he wished to talk about, and could not get back to sleep. He said that he knew his memory was poor, and that if he had waited until the morning he would have forgotten all about the issue that was bothering him. Consequently, he would try to find a staff member to relay his anxious thoughts. Given this understanding, a piece of technology was developed for him. He was asked how he would feel if he were provided with a voice recorder that would enable him to record his concerns during the night rather than go and search for a staff member. He could then replay it to the staff in the morning. He seemed quite happy with this proposal as it meant that the issue would still be dealt with in the morning even though he knew he would have forgotten about it. Care staff felt that normal recorders would be too difficult for him to operate, and after discussion with him a design was constructed that just used one large ‘record’ button on the top. He found this very easy to operate, as he just had to reach over to his bedside cabinet, press the button, and say what was bothering him. Several messages could be recorded.

The device was partially successful. He was able to use it, and was happy that it was an effective substitute for finding care staff to talk to in the night. Unfortunately, the messages that he recorded were not very coherent, and it was difficult for care staff to understand what was bothering him. Although he could not remember what the issue was by the morning, he realized that care staff were not clear about what was bothering him, and this decreased his satisfaction with this method of reporting his concerns. Therefore the device was not completely successful, but the experience is included to give an insight into both the potential of simple technological interventions once a clear understanding of the problem is known, and the need for close and careful involvement of the user in any design solutions.

Interactions between technology and people

Behaviour monitoring

A key aim of the use of sensors is to enable the house to make judgements about the behaviour of the occupants. These behavioural judgements can in turn be used to initiate the action of a support device, or to call for assistance. For example, the bed-occupancy sensors can be used to make a judgement about whether someone is in bed or not, and then turn on lights accordingly. Movement sensors can be used to make a judgement about whether the occupant is showing signs of restlessness, and then initiate a voice prompt to encourage the occupant to go back to bed. Such judgements are relatively easy for human carers to make. For personal carers there is an understanding of the person’s behaviour from a history of living with the person. The carer would know the particular circumstances occurring at any given time that might be affecting behaviour, and, of course, carers have that crucial human capability of empathy, of being able to intimately relate to how the person is feeling at any given moment. Consequently, the judgement of human carers about when to provide assistance, and what kind of help to give, are based on a very sophisticated understanding of the user’s requirements. (Of course, it is common experience that some people are far more skilful in this area than others.) Trying to emulate such approaches through the use of technology is bound to be very difficult, and quite crude in its application. The technology can mostly only respond according to the algorithms, or programme plans, built into it. These, in turn, have been designed according to very simple cause-and-effect rules that reflect the designer’s general understanding of the significance of the information that has been obtained. This information is primarily the sensor data, although it is possible for the system to take past behaviour of the user into account.

Therefore the technology is quite primitive in its ability to make accurate judgements about user’s behaviour, and it is often likely to be inaccurate. Therefore the kind of support that it can initiate is often going to be far from ideal. This is an important issue because so much exploratory work in this field is technology led, and assumes that these judgements are much easier to make and much more accurate than they often are. If judgements are being made about whether someone is in bed or not, these errors would be annoying, and possible confusing, but are not likely to affect the user’s safety. However, if judgements are made about the occupant’s use of a cooker, there is a real risk that the user’s safety could be compromised.

An illustration of the difficulty of making such judgements with cookers concerns a system used to assess whether someone had left a pan or kettle on too long, and it had boiled dry. The technology was based around simple infra-red sensors. These sensors looked at the side of the pan, and could provide information about the temperature of the pan from the infra-red light that is emitted. The pan temperatures monitored can determine within a few seconds if a pan has boiled dry. Based on these observations a simple sensor system was developed that was mounted at the side of a cooker and looked at one spot on the side of the pan on a cooker ring (Figure 13.4). Figure 13.5 shows the kind of data that such a sensor collected. As can be seen the signal increased as the pan was being heated to boiling, then stayed the same during boiling, and then rapidly increased when the pan boiled dry. It was hoped that the rapid increase when the pan boiled dry could be detected, and used to turn off the ring.

Fig. 13.4 Pan-boiling-dry sensor mounted on the side of a cooker.

Fig. 13.4
Pan-boiling-dry sensor mounted on the side of a cooker.

Fig. 13.5 Raw data from a sensor showing the time course of pan temperature as the pan boiled dry.

Fig. 13.5
Raw data from a sensor showing the time course of pan temperature as the pan boiled dry.

Unfortunately, the behaviour of people is more complicated than this simple model assumes. When cooking, people will often turn the heat up and down. They will move pans around between rings. These and other behaviours make it very difficult to use the raw sensor data to make a judgement about what is going on. In order to develop appropriate control algorithms it is important in these situations to collect a lot of data from real-life usage of cookers to see how people react, and to try and design the algorithm to accommodate the variability that occur. Having followed this procedure, it was anticipated that the algorithm could make a correct judgement about 80 per cent of the time. This was a high rate, but still meant that in 20 per cent of the cases it was making wrong judgements. The next step was to allow falsepositive decisions, where the cooker turned itself off even though there was no danger, but this proved to be very irritating to users. In such a situation it might be possible to improve the success rate by incorporating learning capabilities in the control software so that the cooker adapted to the kind of behaviour displayed by a particular user, but even then the success rate would not be 100%. It has to be accepted that the judgements being made when human behaviour is involved are probabilistic and that there will be errors. Consequently, it is very important that some form of back-up is provided to deal with these errors. In the case of the cooker it was decided that if any of the sensors continued to indicate danger for more than a short time after the cooker had acted to prevent it, the house would turn the whole cooker off, or shut off the gas if it was a gas cooker, and then call for some assistance. The helper would then have to check that everything was in order before turning the cooker on again. Such back-up facilities are very important when dealing with the variability of human behaviour, and safety critical judgements are being made.

User interactions

Most of the technology discussed does not require much in the way of direct user interaction. It has a monitoring role, and it is linked to simple support devices that are brought into play in an autonomous fashion. This is the beauty of smart home technology. Its operation is not dependent on the executive skills of the user. However, there is still a need for some interaction and choice on the part of the user. Interactions can be quite simple. If a cooker monitor detects that excessive smoke is being generated by hot fat and turns the cooker off, it needs to provide the user with some feedback and the opportunity to resolve the situation. The feedback can probably best be done using voice information, but if the user is given the opportunity to turn the system back on there will inevitable be some interaction and control that the user has to follow. The same requirement can arise from specialist devices such as the night-time message system described above. The user had to initiate the recording facility. If such interactions are going to be done confidently, and effectively, by the user, some thought has to be given to the nature of such controls. A useful rule of thumb for users with executive problems is to make controls look and feel familiar to ones experienced before. Therefore no learning is involved.

The approach of maintaining familiarity works quite well, but this is not always possible. However, evidence is emerging that for all such users it is possible to develop controls that are very intuitive to use. For example, a simple music player was developed that enabled someone with cognitive problems to turn it both on and off and to select the kind of recorded music they wished to listen to (Figure 13.6) (Orpwood et al. 2007; Sixsmith et al. 2007). The controls typically fitted to recorded music players, with their multitude of small buttons and little lights, would be quite inappropriate for these users. The simple music player stored many recordings, like an MP3 player. It was turned on and off by lifting and shutting its lid, like an old musical box. Inside was a large illuminated and raised button decorated with icons that communicated ‘music’. When the user pressed the button the player changed the kind of music it was playing, like changing a CD. Users just pressed the button until something arose that they wanted to listen to at that time. These simple controls worked well, but they were far more than just a good idea on the part of the designer. A lot of time was spent with potential users exploring in an iterative manner how they reacted to different kinds of controls, and evolving systems that were intuitive. The design of such controls requires a lot of work with users, and certainly is not something that can be done successfully from a third-party understanding of the issues.

Fig. 13.6 Music player designed to be very intuitive to operate by someone with cognitive problems.

Fig. 13.6
Music player designed to be very intuitive to operate by someone with cognitive problems.

Future

The technology used for autonomous homes is developing all the time, and this development is being directed by the accumulating experience of what works and what still causes problems. It is clear that wireless-based sensors and support devices will continue to evolve further until a truly comprehensive range of equipment is available. The power management of these devices is also likely to improve with time. It is also likely that adaptive control systems will be used more. Such systems learn from their experiences, and so adapt themselves more closely to the behaviour and the needs of the individual (e.g. Mozer 2004). A number of such systems are being investigated, and these are likely to become more mainstream. The link between technology and service provision reqires much research, and this area will surely see many improvements. It is important to ensure that care plans can easily be adapted to incorporate the facilities the technology can provide, and for the technology to link into the kind of clinical information required for revision of care plans. There is a need for the information collected from the sensors to be made available in an easily understandable form to give care-providers more insight into how the user is getting on. But, most importantly for the future of this technology, there is a real need to gain far more experience of how users react to these systems. What are the main benefits? What are the problems? Confident adoption of the technology will only take place once this evidence is available and confirmed by different evaluation programmes.

Conclusions

The use of autonomous smart environments for people with executive problems is still in its infancy and many issues need exploring. However, there is no doubt that the potential of the technology has been demonstrated. It can provide support alongside human care and augment the assistance that support staff can provide. Its future development and the work needed to realize its potential depends very much on ensuring such developments are user led. Its development has to reflect a growing understanding of how people can interact with the technology, and how its support can be applied most effectively. Given these important provisos, the use of smart technology has an important future in the care of people with executive disorders.

References

Bjørneby, S. (1997). The BESTA flats in Tonsberg. Using technology for people with dementia. Human Factors Solutions, Oslo.Find this resource:

    Dewsbury, G., Clarke, K., Rouncefield, M., Sommerville, I., Taylor, B., and Edge, M. (2004). Designing acceptable ‘smart’ home technology to support people in the home. Technology and Disability, 15, 191–201.Find this resource:

      Evans, N., Orpwood, R., Adlam, T., Chadd, J., and Self, D. (2007). Evaluation of an enabling smart flat for people with dementia. Journal of Dementia Care, 15, 33–6.Find this resource:

        Gann, D., Barlow, J., and Venables, T. (1999). Digital futures: making homes smarter. Chartered Institute of Housing, Coventry.Find this resource:

          Gibbs, C., Adlam, T., Faulkner, R., and Orpwood, R. (2003). Development of a cooker monitor for people with dementia. In Assistive technology: shaping the future (ed G.M. Craddock, L.P. McCormack, R.B. Reilly, and H.T.P. Knops), pp. 771–5. IOS Press, Amsterdam.Find this resource:

            Hanson, J., Osipovic, D., Hine, N., Amaral, T., Curry, R., and Barlow, J. (2007). Lifestyle monitoring as a predictive tool in telecare. Journal of Telemedecine and Telecare, 13 (Suppl 1), 26–8.Find this resource:

            Jepson, J. and Orpwood, R. (2000). The use of SMART home technology in dementia care. Proceedings of the Annual Conference of the College of Occupational Therapy.Find this resource:

              McKenna, S.J. and Nait Charif, H. (2005). Summarising contextual activity and detecting unusual inactivity in a supportive home environment. Pattern Analysis and Applications, 7, 386–401.Find this resource:

              Mozer, M.C. (2004). Lessons from an adaptive house. In: Smart environments: technology, protocols, and applications (ed D. Cook and S. Sas), pp. 273–94. John Wiley, Hoboken, NJ.Find this resource:

                Orpwood, R. (2006). Smart homes. In: Principles and practice of geriatric medicine (4th edn) (ed M. Pathy, J. Morley, and A. Sinclair), pp. 189–198. John Wiley: Chichester.Find this resource:

                  Orpwood, R., Gibbs, C., Adlam, T., Faulkner, R., and Meegahawatte, D. (2004). The Gloucester Smart House for people with dementia: user-interface aspects. In: Designing a more inclusive world (ed S. Keates, J. Clarkson, P. Langdon, and P. Robinson), pp. 237–245. Springer-Verlag, London.Find this resource:

                  Orpwood, R., Gibbs, C., Adlam, T., Faulkner, R., and Meegahawatte, D. (2005). The design of smart homes for people with dementia: user-interface aspects. Universal Access in the Information Society, 4, 156–64.Find this resource:

                  Orpwood, R., Sixsmith, A., Torrington, J., Chadd, J., Gibson, G., and Chalfont, G. (2007). Designing technology to support quality of life of people with dementia. Technology and Disability, 19, 103–13.Find this resource:

                    Perry, M., Dowdall, A., Lines, L., and Hone, K. (2004). Multimodal and ubiquitous computing systems: supporting independent-living older users. IEEE Transactions on Information Technology in Biomedicine, 8, 258–70.Find this resource:

                    Porteus, J. and Brownsell, S.J. (2000). Using Telecare: exploring technologies for independent living for older people. Anchor Trust, Oxford.Find this resource:

                      Sixsmith, A., Orpwood, R., and Torrington, J. (2007). Quality of life technologies for people with dementia. Topics in Geriatric Rehabilitation, 23, 85–93.Find this resource: