Group Robot Intervention on Paediatric Inpatients and Their Caregivers, Using 'New aibo' by Sony

The study on robot-assisted therapy in a pediatric eld has not been applied suciently in clinical settings. The purpose of this study is to explore the potential therapeutic effects of a group robot intervention (GRI), using dog-like social robot (SR) “aibo” in pediatric ward. GRI by aibo was conducted for those children with chronic illness (127 in total) who are hospitalized in National Center for Child Health and Development (NCCHD), and their caregivers (116 in total), from March to April 2018. The observer made structured behavioural observation records, based on which qualitative research on the features of their words and conducts were carried out. As a result, rst, during the GRI, about 2/3 of total expression by children were positive, while about 1/4 were negative or inappropriate. On the other hand, as seen in the “change” group, those children who had originally responded with negative expression eventually came to express positive expression, while getting involved in a ternary relationship or participating in a session more than once. Secondly, as for the expression from the caregivers during the GRI, active expressions such as “participation” and “exploration” accounted for the 2/3, while 1/3 turned out to be rather placid expressions such as “watch over” or “encourage. ” Conclusion: There has not been any precedent study on the features of words and conducts expressed by patients and their caregivers during the GRI by aibo. The outcome suggests that aibo could possibly be used as a tool for group robot-assisted therapy in the pediatric treatment setting.


Introduction
Hospitalization is a painful experience for children. 1 Becoming ill and hospitalized, children are disconnected from their familiar lives and lose their social ties. 2 These experiences could lead to feelings of loss of control and negative effects on a child's natural development and their physical and emotional health. 3 Although psychosocial support provided by health professionals such as child life specialists or nurses is necessary, it is limited in terms of human resources. [4][5] In this context, Animal Assisted Therapy, 6 as a complementary and alternative treatment for pediatric inpatient settings, has already been implemented in some pediatric hospitals. Animal Assisted Therapy, however, has risks of infection, injury, bites or animal allergies. Hence the robot-assisted therapy had been developed and tested in elderly population, in which PARO, a baby seal-like social robot (SR), was offered to people with dementia. The result showed improvement in stress and anxiety as well as better communication skills. [7][8] Still, the study on robot-assisted therapy in a pediatric eld, has only just begun. 3,[9][10][11] Though as many as 26 different kinds of SR have been reportedly used so far, 2 none has yet to be applied in clinical settings.
"New aibo ( Figure 1) (aibo, hereafter), " an Arti cial Intelligence (AI) dog-like SR for household use, was released from Sony in 2018. aibo has distinctive characteristics of comforting effects due to its round dog-like appearance, interactive communication skills, and growth in response to human interactions. A study has been reported in which an old type AIBO was implemented into pediatric inpatient setting where the interaction between robots and humans was examined. 12 As for aibo, however, no study has ever been reported in the pediatric nor adult setting.
As a household product, aibo has distinctive traits of such as 1) comforting effect, 2) interactive communication skills, 3) growth in response to human reaction. It suggests that aibo may hold some therapeutic effects toward pediatric patients as well as their caregivers in the inpatient treatment setting.
In order to answer this clinical question, we designed a pilot study to explore aibo's potential effects. We conducted a robot-assisted intervention by aibo for the pediatric patients with chronic illness (127 in total) who are hospitalized in the National Center for Child Health and Development (NCCHD) and their caregivers (116 in total). Based on its result, we carried out a qualitative research on the extracted features of words and conducts expressed by the patients and caregivers.
Materials And Methods 1. Study Design: This pilot study is a prospective observational study which carries out a hypothetical search based, qualitative evaluation.
2. Participants: 127 children (0y 6 months to 13 years old) who were hospitalized in the Internal Medicine and Surgical wards of the NCCHD, and 116 caregivers were assembled as participants. The number of children who participated in single intervention was 1 to 15 (median 4.5), caregivers being 1 to 10 (median 4.5). The inclusion criteria was children who have been hospitalized for more than 48 hours.
The exclusion criteria was children who were not considered to be well enough to play in the playroom, such as those who had infection, who were compromised, who needed rest, and who were medically at risk.
3. aibo: aibo, a household use SR released by Sony, is equipped with AI and has the following traits; for instance, Dog-like appearance and behaviour: A rounded shape that gives a sense of warmth, lively form with cuddly sweetness. It tries to interact with you, speaking with its eyes, making eye contact with you. Vivid, dynamic movement with a variety of lovely gestures.
Communication: Curious aibo possesses feelings and desires, based on which it decides what to do, sometimes acting in unexpected ways. Through its eyes (camera) that recognizes the person or the place, aibo understands the people around it and changes its behaviour. It detects obstacles, steps or people, against which it tries to avoid or decides what to do. aibo has got ears to understand human words and is able to change its behaviour. It reacts to the human voice, perceiving where the sound comes from. aibo shows pleasure, when it is praised and stroked on those sensors on its head, chin and back.
Growth: aibo could turn into a spoiled or a wild one, depending on the way people interact with. The more frequent it sees a particular person, it remembers that person, recognizing his/her face. If you teach aibo a funny pose or a dance, it is able to learn the new movement.

GRI by aibo
From March 14 to April 12 2018, 24 series of GRI were conducted for 30 minutes at a time during 14:00 to 15:30 in the playrooms of nine medical wards of Internal Medicine and Surgery, at NCCHD. One child psychiatrist or a clinical psychologist attended as an assistant and an observer (A/O). The size of the ward playroom is about 3 by 5 meters. At the xed time, the A/O turned on aibo to begin the GRI. The A/O saw to it to ensure that no harm would be caused upon children, caregivers and aibo. At every session, the A/O explained to children and caregivers aibo's basic movements, how to touch aibo to please him, and how to bring out songs, poses or dances that aibo expresses. The A/O observed the children and caregivers as they naturally played with aibo, and recorded their words and behaviour on the behaviour observation records (Figure 2). At the end of 30 minutes, the A/O made a statement such as "It's time for aibo to go to bed" and turned off aibo, stroking and wiping it with clean cloth to nish.

Qualitative examination
The 24 behavioural observation records obtained from 24 series of GRI, were sorted by sentences into verbal / non-verbal expressions by children verbal / non-verbal expressions by caregivers. Each sentence obtained was written out on a sheet, and named for relevant groups to be integrated. 13

Ethical Consideration
In this study, although no informed consent had been obtained from the participants, due to the nature of the study, su cient consideration was given so as not to identify a particular patient. Approval has been given by the Ethics Committee in NCCHD. Table 1 shows the date and time, number of children and caregivers who participated in each session. 24 behavioural observation records were collected from the 24 GRI sessions. Out of the 24 records, sentences regarding verbal / non-verbal communication by children and the caregivers were sorted and extracted. There were 140 sentences on children, and 43 on caregivers. The sentences on children were sorted according to a qualitative point of view into the following 9 categories: 1.'a nitive / positive / cooperative', 2.'empathy / comforting / healing', 3.'defensive / cautious', 4.'fear / anxiety', 5. 'dependent / dominant / one-sided', 6.'rebellious / aggressive', 7.'indifferent', 8.'rejective / crying', and 9.'change. ' The examples of sentences assigned to each category are shown in Table 2. Figure 3 shows the breakdown of children's words according to the 9 categories. Among the expressions by the children obtained from the GRI by aibo, positive expressions such as 1.'a nitive / positive / cooperative', 2.'empathy / comforting / healing' accounted for roughly 2/3 of the total. On the other hand, about 1/4 of the total responses were negative or inappropriate such as 3.'defensive / cautious', 4.'fear / anxiety', 5.'dependent / dominant / one-sided', 6.'rebellious / aggressive', 7.'indifferent', 8.'rejective / crying'. In addition, 9.'change' accounted for about 1/10, which indicated those children who had originally responded negatively eventually came to express positive reaction, through observing the behaviours of others, getting involved with others, or participating in the session more than once. Figure 4 shows the result from ternary relationship, namely, "human-aibo-human" interaction. Of the positive expressions, ternary relationships in which a caregiver, the other child and A/O shared a pleasure were detected at a high proportion. In the category 'change' , in particular, the ratio of the ternary relationship was high. Next the sentences on caregivers were sorted according to a qualitative point of view into the following 4 categories: 1.'participation', 2.'exploration', 3.'watch over', and 4.'encourage'. The examples of sentences assigned to each category are shown in Table 3. Figure 5 shows the breakdown of caregivers' words according to the 4 categories. Positive reactions such as 1.'participation' and 2.'exploration' accounted for 2/3 of the total, while placid reactions such as 3.'watch over', 4.'encourage' accounted for 1/3.

Discussion
This study is the rst survey to report a qualitative research which was carried out on the extracted traits of words and behaviour expressed by the patients and their caregivers, when aibo was used for a GRI in the pediatric ward. Two points were identi ed. First, positive expression by children accounted for as much as 2/3 of the total. Besides, the children from 'change' group originally responded negatively but eventually came to express positive reaction, through getting involved in ternary relationship, or gradually getting accustomed to the session by participating more than once. Secondly, among the expressions by the caregivers obtained from the GRI by aibo, positive reactions such as 'participation' and 'exploration' accounted for 2/3 of the total.
First of all, in the GRI by aibo, positive expression by children accounted for as much as 2/3 of the total. Besides, the children from 'change' group originally responded negatively but eventually came to express positive reaction, through getting involved in ternary relationship, or gradually getting accustomed to the session by participating more than once. In a recent study, Haggable, a bear-shaped SR, Haggable on a tablet screen, and a bear shaped toy were offered to 54 hospitalized children as an interventional therapy.
Video monitoring analysis showed the children of 'SR' group expressed more of "joy" or "comfort" , than of "sadness", 5 which is similar to the study result we obtained. In another study, Nao, a humanoid SR, was implemented for hospitalized children, for the purpose of pain reduction as well as increasing joy and motivation for treatment. Although the number of cases were small, the study reported that Nao helped the children reduce their anxiety, anger and depression, playing a role as a good friend to encourage them. [14][15] Becoming ill and hospitalized, children are disconnected from their familiar lives and lose their social ties. 2 In most cases, children may lose the precious opportunities to build ternary relationship, cooperative attention, imitation and so forth, 16 all of which encourage the development of children. In our study, it is signi cant to point out that a certain number of children have experienced a scene to change their feeling into that of positive, through building ternary relationships, as seen in the 'change' group. It suggests that aibo could possibly be useful as a tool to help build ternary relationships that enhances sociality of children, which is often impaired in pediatric chronic hospitalization. Although there are various studies which examine the potential use of SRs in pediatric settings, the number of participants is small. 2 aibo was implemented into an inpatient treatment setting, whose effects were evaluated for 127 population. No such study has ever been reported in the pediatric nor adult setting.
Secondly, as for the expression by the caregivers in the GRI by aibo, positive reactions such as 'participation' and 'exploration' accounted for 2/3 of the total, while placid reactions such as 'watch over', 'encourage' accounted for 1/3. In the study in which PARO was implemented into pediatric inpatient setting, it was reported that when family members participated together, the higher the rate of improvement on a child's anxiety and pain. 10 In our study as well, it was potentially signi cant for the children to see their caregivers participate positively. Furthermore, the caregivers of children with chronic illness tend to be more susceptible to mental distress such as anxiety or depression than the caregivers of healthy children. 17 From caregivers' mental health perspective, in this study, it was noteworthy for the caregivers to have enjoyed themselves 'participating' in the aibo intervention, or to have communication with other caregivers to share the excitement into 'exploration.' aibo is bene cial in making up for human resources as well as from the nancial perspective. Moreover, for many pediatric patients, child life specialists are helpful in reducing the pain experienced during hospitalization. 5 Service provided by human labor, however, has limits. For there is not enough child life specialist staff to offer help for every child in every opportunity experienced during hospitalization. SRs, including aibo, could ll the gap between the number of pediatric inpatient children in need of psychosocial care and the lack of human resources to support them.
Three aspects of limitation of study should be noted as shortcomings. First, in this study, age and gender of the participants were not accurately obtained. Secondly, neither multilateral, scienti c nor quantitative evaluation was not made. Third, this report had no control group.
There are two ndings from this world's rst study which examined the potential effect of aibo onto the patients in inpatient treatment setting. Substantial expressions by the children obtained from the GRI by aibo were positive, while a certain number of children who had originally responded negatively eventually came to express positive reaction, through building ternary relationships or gradually getting accustomed to the session by participating more than once, as seen in the 'change' group. Whereas much of the expressions by the caregivers obtained from the GRI by aibo were positive reactions such as 'participation' and 'exploration.'

Conclusion
This study is the world's rst survey and report on the features of words and behavior expressed by the patients and their caregivers, when aibo was used for a GRI in the pediatric ward. The outcome suggests that aibo could possibly be useful as a tool for GRI in pediatric treatment setting. When conducting a robot-assisted therapy with aibo in the future, we plan to carry out a multilateral, scienti c as well as quantitative evaluation with a biopsychosocial perspective. The date that support the ndings of this study are available from the corresponding author on reasonable request.

Not applicable
Authors' contributions Dr. Makino and Dr. Tanaka conceptualized and designed the study, collected data, conducted the analysis, drafted the initial manuscript, and reviewed and revised the manuscript. Dr. Tanaka supervised the analysis and interpretation, and critically revised the manuscript for important intellectual content. Dr. K. Nakamura and Ms. Hayakawa contributed to the study design and interpretation of data, and reviewed and revised the manuscript. Dr. A. Nakamura and Dr. Uchida contributed to the study design, and reviewed and revised the manuscript. Dr. Kasahara contributed to the conception and coordinating the project, and reviewed and revised the manuscript. Dr. Kato and Dr. Igarashi contributed to the conception and the planning the project, and reviewed and revised the manuscript. All authors approved the nal manuscript as submitted and agree to be accountable for all aspects of the work.

Ethics approval
Approval was given by the Ethics Committee in NCCHD.

Consent to participate
No informed consent had been obtained from the participants due to the nature of the study. However, su cient consideration was in place so as not to identify any particular patients.

Consent for publication
Not applicable Tries to associate with aibo, actively from the beginning.
Calls out happily, "Ah! Here comes aibo!" "Bye bye, Thank you." Making eye contacts with his mother happily, looking at what aibo does.
An elder child is gently encouraging a younger child who seems nervous.
"aibo's gone to sleep." All are looking satis ed.

3.Defensive / Cautious
First session to participate. Looking at aibo from afar.
Holding hands with her grandfather, observing aibo from afar.
Touching aibo by using another doll as a mediator.

4.Fear / Anxiety
Two infants were timidly watching aibo from afar.
Timidly trying to touch aibo from behind.
Looking at aibo, the child hesitated and turned to the observer.

5.Dependent /Dominant/ Onesided
Not able to wait for aibo to move in a slow action, try to make aibo move as she wishes. Tries to take aibo away from another child.
Tries to keep aibo to herself. Tries to touch aibo, telling "Paw, paw."

6.Rebellious / Aggressive
Could not keep aibo to himself, said nasty words, got angry and left.
Get angry at aibo for not turning around or not responding.
When aibo's battery was out, all said in unison "Dead!Dead!" 7.Indifferent Engaging in different activity near aibo.
Moved toward a table to play board game.
Looking at aibo from afar, seemingly not interested.

8.Rejective / Crying
When an observer or a nurse brings aibo closer, the child fears and cries out and did not stop crying at every trial.
When the baby reached out for aibo, aibo barked and the baby cried. People around the baby were appeased.

9.Change
At rst looking at aibo from afar, then with relief, gradually come closer happily.
At rst looking from afar, then seems relieved to see other children and adults touching and talking to aibo, begins to touch.
Though scared at the 1 st session, after participating more sessions, gradually got used to aibo and began to touch. Associating with aibo, together with the child.
Excited together to see the interaction between the child and aibo.

2.Exploration
Taking photos of the interaction between the child and aibo with a smartphone.
Having conversation with other parents, looking at the interaction between the child and aibo from afar.
Interaction between parents and grandparents are built, Getting to know and Exchanging words with one another.
3.Watch over Watch over the child and aibo in association, from afar.
Reserving her/himself from the child, watching over from afar.
Looking from afar, smiling.

4.Encourage
Encourage the child to associate with aibo.
Encourage the angry child to try touching aibo again.
Warn the child not to handle aibo roughly. Figure 1 aibo    The percentage of ternary relationship, namely, "human-aibo-human" interaction.