Best Practice Guidance
Human Interaction with Technology in Dementia

Guidance

More high quality, ecologically valid, controlled studies must be planned, funded and executed in order to properly evaluate the effectiveness of technologies designed to be used by people with dementia and to improve social participation and self-management.

Explanation and examples

A systematic review found that in the whole world only nine controlled evaluation studies with technologies designed for people with dementia have been carried out in ecologically valid settings, to assess effectiveness in improving social participation and self-management. Controlled studies are the most effective way of conducting unbiased evaluations, from which causal inference can be drawn. Policy-makers should be demanding this level of evidence as a condition of investment in such technologies. So far, studies have been conducted with VR-based technologies, other wearable technologies, and software applications. However, only a single study was found to be of good quality. Other technologies for people with dementia have not yet been the subject of a single ecologically-valid, controlled study with these outcomes (this includes, for example, social robots). In order to conduct high quality studies, researchers must ensure that studies are adequately statistically powered based on a sufficiently large sample; include active technology-based control interventions, so that is controlled for attention; and conduct and report intention-to-treat analyses, taking into account data of all participant to the study, including dropouts, and not only those who completed the intervention. Funding bodies must recognize the need to fund such studies accordingly. Clinicians, healthcare providers, policymakers and users of technology should expect and demand that such high-quality evidence is available to support decision-making.

Guidance

Ensure that the methodology for the evaluation of social robots for older adults with and without dementia is appropriate for the purpose of the study, to strengthen the results of the study.

Explanation and examples

Social robots are seen promising for supporting daily functioning and promoting overall social health of cognitively impaired older people, particularly those with dementia. Our scoping review into methodologies used to study the feasibility, usability, efficacy, and effectiveness of social robots for elderly adults with and without dementia showed that, despite promising results, the quality of studies remains low due to various methodological limitations. We have therefore formulated recommendations focusing on different types of studies that can help future researchers develop appropriate study designs to evaluate social robots, allowing for more reliable information on study outcomes:

• For feasibility and usability studies an experimental design with mixed-methods of data collection (qualitative and quantitative) are recommended. Multiple interaction sessions with the social robot are recommended as they may reveal changes in feasibility and usability, when the novelty effect gradually fades and people get used to the robot.
• Appropriate designs for efficacy and effectiveness studies are RCTs, or quasi-experimental designs when randomization is not feasible. Sample sizes should be sufficiently large, and individual interaction sessions with the social robot running for more than one month would serve best for such studies to obtain relatively robust and reliable results. Efficacy and effectiveness should only be studied in fully functioning social robots.
• It is strongly recommended not to combine different aims in one study. The preferred designs to study the feasibility and usability of a social robot, differ significantly from the designs needed to study efficacy or effectiveness.