Our study found the RNA was generally perceived in a positive light. Patients engaged with the RNA to complete functions and found its anthropomorphic features pleasing. Nurses saw the potential of the RNA freeing up their time so they could provide more direct patient care. However, we found a few technical challenges; the RNA did not meet expectations in performing vital signs monitoring, medication delivery did not guarantee reliability, and RNA’s movement could be faster. Patients and nurses suggested to include social interaction function to enhance RNA’s role in the ward. The RNA’s ability to save time would require further investigation.

We found patients had positive interactions with the RNA. Patients smiled at the RNA and thanked it for its service as they would to a human being. RNA’s female voice and human-like torso might have conjured an image of a nurse, which allowed patients to relate with the RNA more comfortably [16, 17]. We suspect the ubiquity of robots in daily lives in Singapore, such as cleaning robots in malls, and the frequent depiction of robots in popular culture, also contributed to a sense of familiarity to its existence and a high level of acceptance [18].

Patients were generally satisfied with the RNA’s performance across the three tasks. They indicated good user experiences despite experiencing errors and having to play a more active role. They were most comfortable with the simpler function of medication and item delivery. For more elaborate processes of vital signs measurement, the RNA’s instructions were not as well understood and more than half of the patients made multiple attempts using the oximeter and voice-input pain score. Even after experiencing these challenges, only a quarter attributed it to insufficient instructions provided. Patients attributed the difficulties to “patient-as-user-issues”, “lack of familiarity”, and perceived the RNA to function well in their recount of the trials. Patients’ willingness to accommodate the RNA could be their lowered expectation that the RNA was not a perfect replacement for nurses [19]. Imperfections in social robots have shown to increase likability as they appear less distant [20].

Patients desired more social interaction capabilities beyond basic greetings. They suggested more two-way conversations, to better replicate how a nurse would communicate with patients. This feedback highlights the multifaceted nature of nursing care—going beyond clinical tasks and the importance of human interaction in the nurse-patient relationship, which is an area of ongoing and future development [21,22,23].

Patient surveys and nurse focus group discussions revealed the concept of time efficiency is not straightforward and has different implications for different stakeholders. Patients experienced long wait times because the RNA is programmed to move slowly and sometimes required multiple attempts at their tasks. This could disrupt patient’s rest and recovery. For nurses, it was perceived to be acceptable that the RNA took longer to perform, in particular lower value tasks, such as item delivery. However, for higher value tasks, such as medication delivery, the trade-off between speed and reliability would need to be considered more carefully as it has serious implications.

Time efficiency is further complicated by the value which task substitution adds. While studies in some sectors viewed robotics more favourably if it improved task efficiency, accuracy, consistency, and speed [24, 25], other studies perceived robots to be valuable when substituting humans in time-consuming and repetitive tasks even if it took longer to perform [26]. At present, the RNA is unable to independently perform as it relies on nurses to allocate it tasks. If vital signs measurement could be programmed in advance and performed on a fixed schedule, it would improve RNA’s independence. Another approach would be to develop a more intuitive learning system that could facilitate better understanding of patients’ needs and respond more effectively (such as delivering a glass of water). With increase in RNA’s independence, the time freed up could provide nurses take on tasks better performed by humans, such as patient education and emotional support or more time for rest [24, 27]. Our study suggests there is a need to examine whether time efficiency or task substitution adds value to a nurse’s work life. One approach would be to perform a time and motion study of robots carrying out nurse-related tasks to ascertain potential time-savings generated by the RNA and at the same time examine how the time saved is used by nurses [28, 29]. For example, in a study of a pharmacy dispensary robot, researchers found an increase in work productivity and reduced staff movement in the dispensary—this could be explored for the RNA in a ward [30].

Nurses expressed concerns that older adults, who are the main group of patients in a general medical ward, might have difficulties interacting with the RNA. More work is required to examine the role of patient’s physical limitations, cognitive impairment, and cultural factors that could pose as barriers for RNA’s adoption. Customization and better matching of robot to local human context would be required as those who less tech savvy might find it harder to interact with the RNA [31,32,33,34].

This study has several limitations. First, a bias could have been introduced because patients were briefed prior to the trial and agreed to spend a considerable time to test the RNA’s functions. Patients who voluntarily agreed to the trial might have had higher technological acceptance, thus influencing their experience with the RNA. The participant pool was limited to English-speaking patients, patients without mobility constraints and cognizant to use the device. This reduced other scenarios for examining human–robot interactions. Second, each patient only tested the RNA in a one-time supervised trial; if we repeated the trial several times, we might have found additional insights and problems arising from implementing the RNA in an acute ward setting. Third, this study assessed one robot; other RNA prototypes deployed elsewhere might have performed differently.

Our study findings suggest a few areas for further exploration: enhancing social interactivity, improving medication administration, and studying the RNA’s ability to ease nursing workload.

Improving social interactivity could expand the scope and function of the RNA. Future research could examine two-way interactions between the RNA and users, such as reactions to patients’ responses [30]. Information obtained during interactions could be conveyed to nurses to alert them to patients who require attention or learn more about their patients, improving the patient-provider relationship [35, 36].

Nurses are concerned that non-compliant patients may not consume medicines delivered by the RNA. The feasibility of incorporating sensors [36, 37], action-validation functions, and persuasive abilities [38,39,40] into the RNA’s ecosystem to verify medication consumption and encourage medical adherence may be examined.

A detailed study is needed to assess whether robots can substantially ease nurses’ workload [41]. We propose a time and motion study of robots carrying out nurse-related tasks to ascertain time-savings generated by the RNA and simultaneously document how nurses use the time saved by the RNA [29].