A systematic search was conducted of MEDLINE (via PubMed), LIVIVO, PubPsych (via ZPID), and IEEE Xplore Digital Library (via the IEEE website). Trial registrations were searched via the International Clinical Trials Registry Platform (ICTRP). The search terms were derived using the population, intervention, control, outcome (PICO) scheme. Figure 1 illustrates the derivation of the search terms and the database search.

The reference lists of the publications identified by the search were scanned for additional relevant publications. Only publications that reported on robotic interaction and control platforms and described a cocreative development process were included.

The criteria for the selection of publications are described in Textbox 1. According to the question of this scoping review, robotic interaction and control systems that were developed in a nursing context with end users (nurses) and evaluated in practice were included.

The term robotic system used in this paper is based on the ISO 8373:2012 definition. According to this, a robot performs useful tasks autonomously, in the sense that it is able to perform these tasks on the basis of its sensor data without human intervention.

Despite the great progress in robotics, it remains a challenge to synthesize a variety of interaction scenarios (eg, speech, image, text, or movement) in a natural way. Research in human-robot interaction includes both multimodal input signals from humans to robots and multimodal output signals from robots to humans [18]. The aim is to improve the user experience, reduce annoying processes, and promote adoption. The latest research approaches in the field of care also need to be considered in order to translate advances in robotics into practice by developing a natural and adaptive style of interaction [19].

The relevance of a robotic intervention to health care is determined by its structure and services. For example, health services can be provided in the outpatient setting by general practitioners or specialists in nonmedical professions, in the inpatient setting, and as rehabilitation services. In addition to the treatment of diseases, prevention and health promotion are also a focus of the health care system [20,21].

As the end users are explicitly defined as nurses, robotic interventions that are more established in the medical context (eg, surgery) and are subject to different frameworks were excluded. The preliminary research identified a great deal of development of robotic systems and application scenarios for nursing care. However, as the purely technical consideration of application has not yet led to comprehensive implementation, new research approaches such as participation should be brought into focus. Even pure evaluations in cocreative design cannot do justice to the problem and were therefore excluded from this scoping review. The language of publication was not restricted to avoid further reducing the search results of relevant publications.

The characteristics of the included studies are summarized and assessed in Multimedia Appendix 1 [22-24]. The summary and preparation of the data can be described as “data-driven thematic analysis,” in which prominent and recurring themes in the literature are identified, summarized under thematic headings, and subsumed into a higher-order theoretical structure through induction and interpretation [25,26].

The databases were searched manually by PM in December 2023. There were a total of 419 hits, from which 75 duplicates were removed. The remaining 344 hits were reduced to 14 full texts after title and abstract screening. After checking the full texts for eligibility, 3 studies were included. The screening process was repeated with the references of the 3 included studies. However, no further relevant sources could be identified, so 3 studies were finally included (Figure 2). Uncertainties regarding inclusion were discussed with PJ. The bibliographic data of the included publications and the full table of extracted data can be found in Multimedia Appendix 1 [22-24].

All publications included were in English. The studies were conducted in 6 countries; 4 of these countries were in the European Union (EU-28), which was represented by 2 publications [22,23]. Both studies were multinational and were partly conducted in Switzerland and Germany. The third included study was conducted in New Zealand [24].

All included publications were feasibility or user studies [22-24]. All had a cocreative study design. This involves users and stakeholders in the development of the robotic interaction and control platform.

One publication did not provide information about ethics committee approval and did not mention that informed consent was obtained from participants or their carers [23].

In addition to involving end users, cocreative development approaches also require consideration of relevant stakeholders who may interact with the end product and have a legitimate interest in its design. Health care is not an individual service provided by a single professional group, but requires interdisciplinary cooperation between different professions. In the spirit of a stakeholder analysis, the participants in the included publications are identified below.

The 3 included studies reported a total of 299 participants (Table 1), with additional carers including informal carers in one study and both formal and informal carers in another study. In one publication, the term carer could not be assigned to formal or informal care. In addition, one study included family members, facility managers, general practitioners, pharmacists, sociologists, geriatricians, psychologists, and computer specialists. One study involved professional teleassistants in a service center. The studies were designed in such a way that the researchers involved the carers in a cocreative way in the development of application scenarios and interaction concepts. People older than 65 years were included in 2 studies, while in 1 study the average age of the participants was only reported as 80.5 years. Children and adolescents were not reported. One study did not report the health status of the participants. The other 2 reported on people with mild to moderate cognitive impairment. One study defined intact hearing and vision as inclusion criteria (Table 1).

The 3 studies reported on different robotic interaction and control platforms (Textbox 2). All were developed as prototypes and were not commercially available. One study was particularly noteworthy for claiming that the platform that was developed worked for all available service robots. It took into account the dynamic autonomy of the robot and the different digital skills of the target groups. Machine learning could be used to individualize the robot and extend its range of functions [23].

The capabilities and functions of the robotic systems are retrospectively summarized in 4 themes derived from the studies after review (Table 2). Carrying or manipulating objects was described in 1 case. For example, a water bottle was picked up and brought to the user [23]. Reminder functions were described in 2 studies, including for taking medication or for upcoming appointments [22,24]. Monitoring of people was described in 2 studies, including recording of well-being and sleep quality or emergency detection and assessment [22,24]. Communication with third parties via text, audio, or video was described in all scenarios.

Two studies developed an interaction concept consisting of differentiated user interfaces with an adapted range of functions for different end users (people in need of care, carers, and health care professionals) [22,23]. The user interfaces for health care professionals were described as web-based applications in all publications.

Interaction via screens of different sizes using touch gestures was described in all 3 studies. A combination of audio output and written visualization was also chosen as output in all 3 studies. In 2 studies, facial or audio recognition of the target person was used to start the interaction [22,24]. The following requirements were defined for the user interface: a flat menu structures with step-by-step interactions [23]; no foreign or technical language [23]; large and high-contrast colors, buttons, and fonts [22-24]; and the possibility to adjust the volume [22,24].

Margaritini and colleagues [22] divided the cocreation process into 2 phases. In the first, 3-month phase, people with care needs and their carers (formal and informal) used an initial version of the platform. The aim of this phase was to evaluate the response of older people to the use of this new technology. As a result, technical change requests from older people were identified in order to improve the overall experience with the platform and, in particular, the usability of the graphical interface. At the beginning of the study, participants were introduced to the platform by the research team (consisting of psychologists, biomedical engineers, and physiotherapists) and helped to set it up. Monthly evaluation visits were then carried out to collect quantitative and qualitative data. The second phase followed the same participation pattern. The platform was also tested over a period of 3 months.

Mast and colleagues [23] refer to user-centered design in their research approach. A total of 6 user studies were conducted at different stages of the project. Again, 2 main phases can be identified. In the first phase (preparatory studies), mainly user studies and analytical studies were carried out. A needs analysis was carried out with older people and care staff using a mixed methods design. The focus was on the requirements for robotic systems and their integration into the daily life of end users, as well as on the technical requirements. Iterative design was the second phase. After the technical development, the existing prototypes were evaluated with older people, informal carers, and professional teleassistants. For this purpose, a usability study was carried out to evaluate the different user interfaces with representatives of the target group. The results were then used to revise the platform.

The third study, by Tiwari and colleagues [24], had 3 iterative cycles. The first cycle, which defined the framework concepts, involved developing an understanding of the underlying process that the platform was intended to map. This was done by observing nurses at work and then evaluating the information collected. In addition, interviews were conducted with people in need of care, relatives, facility management, nursing staff, doctors, and pharmacists. In the second cycle (the design of the application), prototypes of the application were discussed with computer scientists, sociologists, geriatricians, psychologists, nurses, and doctors. In the final phase of testing and refinement, a usability study was carried out with people in need of care.

The evaluation methods used by the studies (Table 3) collected both quantitative and qualitative data on requirements for robotic interventions, interactions and controls, usability, ease of use, and acceptance, using established methods such as focus groups, think-aloud, and questionnaires. Observational data were collected using video, photographs, and transcripts. Interviews were conducted without information about the use of guidelines. Recordings of the user interfaces of the robotic systems were also used for evaluation. Carers, health professionals, and experts were involved in the evaluation in addition to older people. All publications reported on established evaluation tools.

The interventions were reviewed for their characteristics. Three studies were conducted in the participants’ home environment [22-24]. In 1 study, the interaction between the participants and the robotic systems followed a schedule set by the researchers [24]. In the other 2 studies, participants were free to interact with the system [22,23].

The duration of the intervention, in terms of participants’ exposure to the robotic system or use per person, was reported in all publications. One study reported 2 intervention cycles of 3 months each [22]. The other 2 studies reported durations of 1 hour [23] and 2 hours [24].

The results of the tests were not described in detail in the trials. One publication merely presented the study protocol for its development [22]. The other 2 studies focused on the presentation of the platforms and the incorporation of the results into development; as subprojects of larger studies, only selected results were presented [23,24]. Technical aspects such as feasibility, usability, suitability, functionality, and specific requests for future features were recorded as outcomes. The user perspective was considered in all 3 studies, with usefulness and interaction with the robot being of interest. Two studies described the preceding requirements analysis, which specified functions and requirements that were crucial for acceptance by the target group [23,24]. An evaluation of the cocreative collaboration between the participants was not mentioned in any of the publications.