Measuring the Impact of a Socially Assistive Robot

The objective of the current study is to evaluate the impact of the robot using a cross-over design.

• Recruitment An existing database with facilities, built up in the previous phases of the project, will be used to recruit new participants. Sample Size calculations were done a priori using G*Power, based on an effect size of 0.4 and testing the difference between two dependent means. According to these calculations, a total sample size of 62 would allow us to detect significant differences among the mean values on the primary outcome measures (α =.025 and power =.8).
• Handling Missing Data Since the reason for dropouts might be of great importance for the results of this study, they will be taken into account. Therefore, the Intention To Treat (ITT) approach will be applied.
• Data Analysis

• Quantitative Analysis

Descriptive statistics will be employed to describe the characteristics of the participants. Distribution of data will be assessed by inspecting the histograms and by appropriate tests, e.g. Shapiro-Wilk Test. Data will be visualized using bar charts (categorical variables) and/or box plots (continuous variables). If appropriate, independent t-tests will be used at baseline data to investigate whether or not there are significant differences between both groups. Alternatively, in case of non-normal distributed data, Mann Whitney-U tests will be applied instead.

Active Engagement Evaluation will be performed at baseline, after 3 months, 6 months, 9 months and 12 months in the study with both group A and B. Primary endpoint is the comparison "no robot" vs "robot" at 6 months within that period. Active Engagement will be calculated by comparing the data of the intervention and control group after 6 months and 12 months with counts (the number of times the person shows a certain behavior at a certain time point during the activity) as the outcome variable. Data will be analyzed using a mixed negative binominal regression, with the subjects considered the random factor and the following variables as fixed effects: (1) condition (robot vs no robot), (2) time (baseline, 3 and 6 months) and (3) interaction between condition and time.

Loneliness Evaluation will be performed at baseline, after 3 months, 6 months, 9 months and 12 months in the study with both group A and B. Primary endpoint is the comparison "no robot" vs "robot" at 6 months within that period. Loneliness will be calculated by comparing the scores on the (full) loneliness scale of the intervention group and the control group after 6 and 12 months. Data will be analyzed using a mixed linear regression, with "loneliness" as the outcome variable and the following variables as fixed effects: (1) condition (robot vs no robot), (2) time (baseline, 3 and 6 months) and (3) interaction between condition and time. The subjects in the study are considered the random factor.

Secondary outcomes Evaluation will be performed at baseline, after 3 months, 6 months, 9 months and 12 months in the study with both group A and B. Primary endpoint is the comparison "no robot" vs "robot" at 6 months within that period. For analyzing the data of the secondary outcomes (physical functioning, cognitive functioning and participation), a linear mixed model will be used (similar to the primary outcomes) if data is distributed normally.

Since multiple comparisons are made, a Bonferroni correction will be applied. The Bonferroni-corrected level of statistical significance will be set at α = .025. Data analyses will be conducted by SPSS Statistics software. For the secondary outcomes, statistical significance will be set at a level of p<.05.

2 - Qualitative Analysis

Focus groups will be conducted after 6 and 12 months to receive feedback from the caregivers. Inductive thematic analysis will be used to analyze the data from the focus groups. This will be done according to the guideline designed by Braun and Clarke (2006), which describes different phases in the process of conducting thematic analysis. In the first phase, the researcher becomes familiarized with the data by a thorough (re)reading from the data from the focus groups. Initial ideas that could possibly be of value will be noted down. The second phase consists of the coding process, where interesting features in the data are being labeled and initial codes will be produced. An interpretative analysis of the data occurs and similarities in the codes will be identified, generating potential themes. These themes will be discussed within the research team and further refined in the next phase, where clear definitions and names will be generated. The analysis will be carried out by the corresponding researcher from each test-site. This in order to enable a focus on the different perspectives resulting from the different settings where the study will be conducted.

Inclusion criteria

• Age 65+
• Score range of 19/30 - 25/30 on Montreal Cognitive Assessment (MoCA)
• Ability to walk independently with or without assistive device

Exclusion criteria

• Health conditions that do not allow participation in a physical intervention
• Planned surgery or hospitalisation within 12 months of the trial
• Insufficient language skills
• History of severe neurological (e.g. stroke) or psychiatric (e.g. depression) condition
• Severe vision impairment / hearing loss
• Participation in another study to test a new technology