XWiki - Test

In this section, we will introduce our experimental test setup and motivate our research questions.

== Problem statement and research questions ==

People with dementia face challenges that were not present in their life before being affected by their condition. This is also influenced by their living situation, since they may live in care homes or dementia centers. To ensure that people with dementia stay safe, activities are usually led by a caretaker. However, the staff has limited time available to spend with each individual person. Group activities or caretaker-led activities may affect the feeling of autonomy of the person with dementia. Autonomy is one of the key psychological needs described in self-determination theory, which describes how feeling fulfilled is influenced by how much a person feels in control of their own actions. It is important to the mental well-being of people with dementia to keep a sense of autonomy.

6

With our research, we aim to provide support to Patients with Dementia (PwD) who currently are dependent on their caretakers and significant other to go for walks. This is outlined in more detail in the Objectives section. The presumed baseline is that PwD walks under the supervision of their loved ones or caretakers. To reach our objective and support our claims, we would like to put forward the following research question:

7

8

//Does walking with WAF increase the perceived autonomy of people with dementia, compared to walking under the guidance of a caretaker?//

9

10

To answer this question, we use the following subquestions:

11

12

* Is WAF effective in providing guidance to a person with dementia on a predetermined route?

13

* Does the increase in perceived autonomy outweigh the possible delayed help when a person with dementia gets in trouble on their walk?

== Method ==

To this effect, we would like to invite users to participate in a usability study. In the study, we will assess our research question by performing within-subject experiments. The participants will walk twice randomly ordered: once with a //caretaker// and once with //WAF//. The former will enact the activity of walking in the current setting and the latter will do the equivalent with the robot. Afterward, participants are assessed by means of Likert-scales to measure the changes. The difference in scores between the caretaker and the WAF setting will then be subjected to a Wilcoxon [[(Conover, 1998, p. 250)>>https://www-wiley-com.tudelft.idm.oclc.org/en-nl/Practical+Nonparametric+Statistics%2C+3rd+Edition-p-9780471160687]] test to determine whether the results are statistically significant. If the significance has been demonstrated, then we can reject the null hypothesis: walking with the WAF will not provide an increased feeling of autonomy compared to walking with a caretaker.

=== Participants ===

For our study, we would have liked to invite people with dementia to participate. However, user testing with a vulnerable group such as this one is difficult seeing the current circumstances of the COVID-19 pandemic.

22

For this reason, we decided to design an experiment to be conducted on our fellow students. Students will be briefed on dementia. Then they take on the role of one of the personas and roleplay a person with dementia. Through this roleplay, we aim to simulate a study with people with dementia as accurately as possible in the given situation. We will give them one page of the [[persona handout document>>https://xwiki.ewi.tudelft.nl/xwiki/wiki/sce2022group02/download/Test/WebHome/personas-handouts.pdf?rev=1.1]] to aid them in their roleplay and study character traits that might influence acceptance of WAF.

23

24

=== Experimental design ===

25

26

The main research question can be answered using a within-subject design in order to be able to measure the //increase// per subject. The learning effect will be mitigated because the two juxtaposed settings are vastly different. So, there is no advantage for the participant to be subjected to two conditions serially. In addition to the questionnaire, notes will be taken during the experiment of the observations of the participant's interactions with WAF.

27

28

=== Ethical Approval ===

29

30

Though the official ethical approval was handled by the course staff, we have also filled in the HREC Risk Assessment and Mitigation Plan form. The form can be found [[here>>https://xwiki.ewi.tudelft.nl/xwiki/wiki/sce2022group02/download/Test/WebHome/HREC-checklist-group-2.pdf]]. Before participating in the study, the participants signed the consent form found [[here>>https://xwiki.ewi.tudelft.nl/xwiki/wiki/sce2022group02/download/Test/WebHome/consent-group-2.pdf]]

=== Tasks ===

We break up our experiment into the following tasks:

35

36

* Walking a route

37

* Wandering off during a walk

38

39

Walking a route

40

The person with dementia will walk a route guided by either a caretaker or by WAF.

41

42

Wandering off during a walk

43

A person with dementia may become lost in their current surroundings. They might wander off from the predetermined route. Depending on the study group, either the caretaker or WAF will guide the person back to the route.

=== Measures ===

To measure the effectiveness of our solution, we use a questionnaire. Since people with dementia may experience difficulties with expressing themselves in a conversation, this questionnaire is led by an interviewer [[(Neerincx et al., to appear)>>]]. A caretaker may also be present to help the person with dementia accurately express themselves.

=== Procedure ===

The procedure is that the person with dementia walks with WAF. The PwD is given WAF and the accompanying proximity bracelet. WAF has a pre-programmed route to walk and the PwD is asked to walk with the dog. While they are walking the dog, they are observed by a separate observer, carefully noting down any remarkable events. The emotional state of the PwD is actively monitored by looking at facial expressions. Furthermore, the researchers will actively check how easily the PwD interacts with WAF to check the familiarity. When the PwD is done walking, some small questions will be asked to check the satisfaction, autonomy, and familiarity.

52

53

Functions like when an elderly person falls or when they walk in the wrong direction will not forcibly be tested with PwD as this might create a dangerous situation. These functions will be tested in advance with the use of "Dogfeeding". Other people from the development team will test these functions.

=== Material ===

The items necessary for the tests are the following

* A MiRo + bracelet

* Caretaker

* Questionnaire

* Consent forms

* Persona handouts

* Controlling device

== Results ==

We asked participants about their sense of being in charge, happiness, safety, and trust when walking with WAF and with the caretaker. In the graphs below we show each sense compared between the two guiding agents.

<tr>

</td>

<p align="left">This comparison shows that when walking with a caretaker the participants felt mostly in charge, whereas when walking with WAF the participants felt this less frequently.</p>

</td>

</tr>

<tr>

<td>

</td>

<td>

This comparison shows that both a caretaker or WAF accompanied positive emotions.

</td>

</tr>

<tr>

<td>

</td>

<td>

This comparison shows a significant difference between walking with a caretaker and WAF. A caretaker made the participants feel safe most of the time or prevented any unsafe feelings. WAF conveyed only minimal feelings of safety with participants answering 'at times' most frequently.

</td>

</tr>

<tr>

<td>

</td>

<td>

This comparison shows similar results to the safety comparison. A caretaker is reported to be mostly or completely trustworthy, while WAF is trusted less as participants trusted it 'a little' or 'mostly'.

</td>

</tr>

</table>

We performed the Wilcoxon test on the questions that were asked to the participants where both WAF and the caretaker were evaluated. With the Wilcoxon test, we are able to determine if there is any significant difference between WAF and the caretaker. Unfortunately, due to the low amount of participants, there is no possibility to create a normal distribution and thus causing the p-value to be less reliable.

108

109

The first question is whether the PwD feels in charge while walking with the caretaker or WAF. The resulting z-value was -0.3145. The value of W is 9. The critical value for W at N = 6 (p < .05) is 2. The test concluded that the result for this question is not significant at p < .05. Thus there is no significant difference between either walking with the caretaker or walking with WAF.

110

111

The second question is whether the PwD feels happy while walking with the caretaker or with WAF. 5 out of the 8 test answers showed no difference in either WAF or the caretaker, they gave exactly the same answers. This caused the test answers to be reduced to 3 which is too small to perform the Wilcoxon test on.

112

113

The third question is whether the PwD feels safe while walking with the caretaker or with WAF. This test had a z-value of -2.3664 and a W-value of 0. The result is significant with p < .05. There was a strong preference for walking with the caretaker in regards to feeling safe.

114

115

The final question is whether the PwD trusts the WAF or the caretaker while walking. The z-value resulted in -2.3664 which resulted in the same W-value as the third question: 0. The result was again significant with p < .05. Again there was a strong preference for walking with the caretaker. The caretaker was considered more trustworthy than the WAF.

== Discussion ==

Through this test, we aim to answer the question: "Does walking with WAF increase the perceived autonomy of people with dementia, compared to walking under the guidance of a caretaker?"

120

We see that in the current state of our prototype, this is not the case. The caretaker is preferred when asked if the user feels in charge, feels safe, and feels trustworthy. However, we note that WAF does slightly outperform the caretaker in regard to happiness. We think that further development is needed to accurately gauge if walking with WAF will eventually score equal or better compared to walking with a caretaker.

121

We see flaws in the limited testing setup that was available to us. With only eight participants, it is nearly impossible to draw conclusions from the experiment. Additionally, these participants were actors and not real people with dementia, which may have caused bias in our results. Finally, while we intended to have the participants walk through a hallway accompanied by WAF, we only had a meeting room available to conduct the experiment in.

=== Inclusivity ===

In addition to the questionnaire, we conducted a short interview with our participant who is hard-of-hearing. Their study was conducted without them wearing their hearing aids.

126

We asked them if they thought their experience would differ if they were wearing their hearing aids. Their answer was that the biggest thing was that in a real scenario, the wristband doesn't give you enough information about what to do. It only tells you that you are something wrong. They commented that if they were 83, they would not have enough direction from this, compared to someone who could hear sounds: WAF could direct them back by barking at the user, for example. However, they mention that if you are severely hearing impaired, there is probably nothing you can do other than vibrating the wristband.

127

We asked them how we could make it more inclusive. They mention that the robot is small and not very visible, which might cause problems for people with a sight impairment. People who are not familiar with walking dogs would also not naturally keep looking at WAF. They recommended using a larger robot.

128

129

== Conclusions ==

130

In conclusion, the prototype as-is is not capable of outperforming the benefits of a traditional walk with a caretaker. Additional work is needed to bring the prototype to a state of usability where users walking with MiRo can feel trust, happiness, safety, and being in charge. We would like to conduct a similar experiment with real people with dementia, in a setting that is familiar to them, to properly simulate a realistic use case. In the future, we would increase the number of participants to gain better insights into the usability of our prototype.

131

132

133

Wiki source code of Test