Expectations of AI-supported communication in nurse-family interactions in nursing homes: Navigating ambiguities in role transformation

Abstract

Introduction

Understanding nurses’ expectations and concerns regarding artificial intelligence (AI)-supported communication with families in nursing homes is essential for designing effective digital tools that support meaningful family engagement. This study explores how nurses perceive the integration of AI-supported communication tools in interactions with residents’ family members, with a focus on their expectations and the ambiguities related to changes to their communicative roles.

Method

This qualitative descriptive study employed semi-structured, in-depth, face-to-face interviews with 15 nurses from seven medium-to-large registered nursing homes in Taiwan, selected through purposive sampling. Data collection continued until thematic saturation was achieved. Transcribed interviews were analyzed using qualitative content analysis to identify key themes.

Results

The analysis revealed a central theme: “navigating ambiguities in role transformation,” which reflected nurses’ perceptions of the opportunities and uncertainties involved in incorporating communication into family interactions. Four categories contain these ambiguities: (1) use of AI to replace repetitive communication (n = 15), (2) emotional interaction (n = 13), (3) transparent data sharing (n = 9), and (4) tailoring of personal care plans for families (n = 7).

Conclusions

Although nurses acknowledged the potential of AI to improve communication efficiency, they also expressed concerns about its ability to preserve emotional depth, clinical judgment, and personalized engagement. Addressing these ambiguities is critical for developing AI systems that align with nursing values and practice. This study provides insight to guide the design of digital solutions that are responsive to the realities of nursing home care and support stronger family involvement.

Keywords

AI-supported communication nursing homes digital family interaction programs registered nurses

Introduction

Family involvement plays a critical role in enhancing the well-being of nursing home (NH) residents, their families, and the nursing staff, forming a cornerstone of individualized resident care.¹ Family involvement encompasses a range of behaviors in which family members provide hands-on assistance, take on the role of overseeing and managing care, and provide socioemotional support to residents.^2,3 Family communication with nursing staff to enhance the well-being of residents, including communication in regard to “the little things,” is an important experience for NH residents.⁴ Transparent, accurate, frequent, detailed, and timely communication from the staff is important for families.^5,6 Effective communication between families and nursing staff is integral to person-centered care, improved resident health outcomes, and greater family satisfaction and involvement.^5–8

Although communication is critical in NH settings, it is often compromised by a lack of standardized procedures, obsolete technological systems, and ineffective coordination at the institutional level.^9,10 NH communication centers on long-term, individualized daily care, unlike the episodic communication typical in acute care settings. Most families visit residents only once a week,^2,11 a schedule that was worsened by the COVID-19 pandemic. These limited interactions contribute to information gaps and delays, increasing the risk of misunderstandings and conflict between families and staff. Families have consistently reported difficulties in accessing timely and accurate information and in effectively communicating their concerns to staff.⁹ These challenges are compounded by chronic staffing shortages and heavy workloads, which further constrain staff's ability to maintain regular, detailed communication. As a result, families often feel disconnected from care, which negatively affects their overall experience with the facility.

Providing families with a straightforward way to access information helps to ensure effective communication, reducing confusion and conflicting messages. This, in turn, can foster trusting relationships between staff and families.⁹ As in many other countries,¹² however, NHs in Taiwan often face staffing shortages. As a result, staff members carry heavy workloads, limiting their ability to provide frequent, detailed, and timely communication with families, which has a negative impact on communication effectiveness.

The COVID-19 pandemic and its aftermath exacerbated these issues by intensifying workload pressures and communication barriers and contributing to increased burnout and staffing shortages in NHs.^13–16 One study found that nurses spend 48.4% of their time on communication in NHs, making it the most time-consuming activity for nursing staff.¹⁷ Other studies have shown that enhancing communication between nurses and residents’ families can be achieved through real-time telehealth services.^18,19

Based on the increasing availability of artificial intelligence (AI) technologies and the increasing demand for integrated care,²⁰ AI-supported communication tools, such as telehealth platforms, AI applications (e.g. chatbots), or predictive tools, can offer promising solutions to longstanding communication challenges in NHs. These tools can streamline information exchange and automate routine communication tasks, helping to reduce the communication burden on nurses. This is particularly valuable in long-term care settings, where face-to-face interactions with family members are often limited. By improving efficiency, consistency, and responsiveness, AI-supported communication not only enhances the quality of nurse-family interactions but also allows staff to focus more on direct, value-added care for residents. At the same time, these technologies are reshaping nurses’ communication roles and introducing new responsibilities into the caregiving process.

Although the COVID-19 pandemic was declared over by the WHO in 2023, its impact continues to underscore longstanding challenges in long-term care (LTC) communication. The pandemic highlighted the critical role of AI-supported communication in maintaining connections between LTC facilities and families. As in-person visits were restricted, AI-based tools became essential for sustaining family engagement and continuity in care-related information. These tools have since evolved from temporary solutions into integral components of LTC practice, supporting both relational and operational needs.^21–23 Communication inefficiencies—such as lack of standardization, outdated technology, and coordination issues—have been widely documented in hospital settings.²⁴ However, few studies have explored the extent to which these or similar communication challenges arise in long-term care environments, particularly in relation to expectations and the use of AI-supported communication tools.

NHs in Taiwan operate under the supervision of the Ministry of Health and Welfare and must adhere to national regulations regarding staffing, care standards, and service delivery. These facilities vary in size and resources but typically employ multidisciplinary teams. Among these teams, registered nurses (RNs) often play a central role in both resident care and communication with families. The standard nursing staff-to-resident ratio is 1: 15.²⁵ These facilities typically care for residents with high physical or cognitive dependency and employ registered nurses, nurse aides, and care managers to provide 24-hour care. Family involvement in decision-making and communication is culturally significant in Taiwan, which adds complexity to how digital or AI-supported communication systems might be integrated.

Although previous research has emphasized the time-consuming nature and importance of communication for families and nursing staff (NS), there is limited research on the expectations of the digital nurse-family communication role from the perspective of nursing staff in NHs. Understanding the expectations of the AI-supported communication role of nurses can offer valuable information for designing effective telehealth interventions to enhance communication between families and nurses.

This study aims to explore how NH staff, particularly registered nurses, perceive and anticipate the role of AI-supported communication tools in their interactions with residents’ families and to examine the challenges they associate with the potential transformation of their communicative responsibilities.

Methods

Design

A qualitative descriptive design using in-depth individual interviews was adopted to explore nurses’ perspectives on using AI-supported communication tools in their interactions with residents’ families in NHs.

Setting and participants

Setting

This study used purposive sampling to recruit the study sites and participants.

Purposive sampling was used to select NHs from the 511 registered in Taiwan.²⁵ We targeted medium-to-large NH facilities located within approximately a two-hour travel radius—by car or public transportation—from northern Taiwan. These facilities were selected because they typically accommodate residents with more complex health needs and employ larger nursing teams, which aligns with the objectives of our study. These characteristics make them well-suited to studying the expectations in regard to AI-supported communication. Seven private NHs were selected: One is hospital-affiliated but operated independently, and six are standalone facilities. All are located in rural areas and operate on a self-funded basis (Table 1). The selection of facilities in rural regions also allowed for the exploration of AI-supported communication challenges in settings in which in-person family visits may be less frequent due to geographic limitations.

Table 1.

Characteristics of participating NHs and interviewees.

NH	Number of beds	Number of interviewees	Role(s) of interviewees	Job status
NH-1	49	2	RN	Full time
NH-2	98	2	RN	Full time
NH-3	162	3	RN	Full time
NH-4	151	3	RN	Full time
NH-5	64	2	RN	Full time
NH-6	49	1	RN	Full time
NH-7	47	2	RN	Full time

NH: nursing home; RN: registered nurse.

Participants

To enhance diversity and support the transferability of findings, we employed maximum variation sampling. Participants for this study were registered nurses licensed in Taiwan, consistent with the MeSH (Medical Subject Headings) classification of “Nurses, Registered,” which refers to nurses who have completed the required education and licensing to practice professionally. They were purposively recruited to ensure variation in years of clinical experience and educational backgrounds to capture a broad range of perspectives. All participants were employed full time in NHs registered with the Taiwan Long-Term Care Professional Association.²⁶ Eligibility criteria included (1) being a licensed registered nurse who worked in a qualifying NH, (2) having the ability to communicate in Mandarin or Taiwanese, and (3) providing informed consent to participate. No other nursing roles (e.g. licensed practical nurses, nursing assistants, and advanced practice nurses) were included. Sampling continued until data saturation was reached, defined as the point at which no new categories or themes emerged from the data.²⁷

A total of 15 female nurses participated in this study, with an average age of ∼33 years. Nearly half of the participants had more than 10 years of work experience, and around two-thirds held university degrees. None of the NHs in which these nurses worked had fully automated electronic health record (EHR) systems. Approximately 40% of the participants reported manually recording residents’ medical measurement data, and about 47% were responsible for manually documenting residents’ medication information. More than two-thirds of the nurses expressed experiencing stress when communicating with residents’ families, and roughly two-thirds were dissatisfied with the current quality of communication. Despite these challenges, all participants expressed a positive outlook and interest in using digital tools to improve communication with family members.

Data collection

This study was conducted between May and November 2024. The primary data collection method consisted of face-to-face interviews with participants. During this part of the study, semi-structured interviews were conducted with nurses. The initial question was intentionally broad to encourage participants to share their perspectives on AI-supported communication with residents’ families.²⁸ Subsequent questions were more specific to develop a deeper understanding. The open-ended questions were as follows: (1) Can you describe your communication experiences with residents’ families in NHs, particularly after the COVID-19 pandemic? (2) What topics do you typically discuss with residents’ families? (3) Have you encountered any challenges or specific expectations when communicating with them? (4) What are your expectations regarding the role of digital or AI-supported communication tools in your interactions with residents’ families? and (5) Which aspects of nursing communication do you believe could be effectively supported or replaced by AI-supported communication tools?” The interviews ranged from 30 to 60 minutes, with an average duration of ∼ 45 minutes.

Those who agreed to participate were recruited by the first author, who subsequently contacted them in person for further interviews. The interviews were conducted in places where the participants felt comfortable. The interviews were recorded using a smartphone. The recordings enabled the researcher to check the participants’ consistency while answering the questions. Further, notes were taken during the interviews to record the participants’ facial expressions, gestures, reactions, and comments.²⁸ This approach also allowed researchers to concentrate on taking strategic and focused notes.²⁹ Verbatim transcriptions of the interview data were also conducted.

Data analysis

The data were analyzed using the qualitative content analysis framework proposed by Graneheim and Lundman,³⁰ which facilitates the exploration of manifest content (explicitly stated) and latent content (underlying meanings). This approach enabled a comprehensive understanding of the communication practices of nurses in NHs and their underlying expectations regarding the use of AI-supported communication tools in interactions with residents’ families. In this study, categories represent the manifest content—the concrete, directly observable aspects of the data—formed by grouping similar codes based on shared meanings. These categories were then abstracted into themes, which reflect the latent content, encompassing broader patterns and deeper insight into participants’ experiences. Themes served as overarching interpretations that synthesized the meaning of multiple related categories.

The analysis followed a six-step process conducted collaboratively by the first and second authors, both trained in qualitative research and with substantial experience—2 and 21 years, respectively—in conducting qualitative studies. First, all transcribed interviews were read thoroughly to gain an overall understanding, and relevant meaning units were extracted as direct citations. These meaning units were then condensed to preserve the core meaning while remaining faithful to the original expressions. Next, the condensed meaning units were labeled with codes, which were compared for similarities and differences and then grouped into subcategories. These subcategories were organized into broader categories and then interpreted at a latent level to derive overarching themes.

The first author conducted all interviews and led the initial stages of analysis, including transcript coding and the development of preliminary categories. The second author reviewed the coding framework, contributed to refining categories and themes, and participated in the interpretation of the findings. Themes were developed and refined through multiple rounds of collaborative discussion to ensure clarity, distinctiveness, and consistency. Any discrepancies were resolved through consensus. To enhance the rigor and credibility of the findings, we consulted an external qualitative research expert. This expert provided an independent review and supported the validation of the interpretations, ensuring that the themes were grounded in the data (Figure 1). This process aligns with best practices in qualitative research by promoting transparency, trustworthiness, and methodological rigor through triangulation and collaborative validation.^31,32

Figure 1.

Procedure of data analysis.

Trustworthiness

To enhance the trustworthiness of the study, we applied the four criteria proposed by Lincoln and Guba³²: credibility, dependability, transferability, and confirmability. These principles guided our research process and helped to ensure analytical rigor and trustworthy findings.

Credibility was established through purposive sampling, prolonged engagement, and peer debriefing. The first and second authors brought substantial experience in NH settings—3 and 21 years, respectively—which fostered trust with participants and facilitated deeper engagement during data collection. Peer debriefings with a qualitative research expert were conducted throughout the analysis to validate interpretations and enhance analytical soundness.

Dependability was supported by the use of a semi-structured interview guide across all interviews, ensuring alignment with the study's objectives. All interviews were audio-recorded and transcribed verbatim. The coding process was carried out systematically and collaboratively by the first and second authors, with regular discussions to refine categories and resolve discrepancies, minimizing individual bias.

Confirmability was enhanced through the use of memos and reflexive journals, which documented analytical decisions, researcher reflections, and the rationale for category and theme development. An audit trail was maintained to ensure transparency and traceability of the research process, helping to demonstrate that the findings were data driven rather than shaped by researcher assumptions.

Transferability was supported by providing rich descriptions of the study setting, participant demographics, and data collection procedures. A maximum variation sampling strategy was employed to include nurses with diverse clinical backgrounds and levels of experience from seven rural, medium-to-large, self-funded NHs in Taiwan. This diversity enhances the relevance of findings across different care contexts. To ensure linguistic accuracy, the second author translated selected transcripts from Mandarin/Taiwanese Chinese into English, and a bilingual medical student back-translated them into Chinese.

Ethics statement

This study was approved by the Institutional Review Board (IRB). The authors also obtained permission to conduct this study from the directors of each NH, as NHs in Taiwan do not have IRBs. All the participants who agreed to participate signed an informed consent form prior to the interviews. The purpose and process of the study, its potential risks, the participants’ right to withdraw at any time and refuse to answer questions, and strategies to protect confidentiality also were explained to the nurses.

Results

Meaning of participants’ expectation

The central theme, “navigating ambiguities in role transformation,” was derived from the latent content. It reflects the uncertainty and duality that nurses face as they transition from traditional communication practices to those supported by AI tools. This theme encapsulates the complex interplay between the perceived opportunities and challenges associated with integrating AI into communication with families. Participants described communication with families in the NH as a continuous and demanding responsibility that required attention throughout the day. The main expectation of AI-supported communication was to replace repetitive and unprofessional but necessary communication. Participants also were concerned that they needed to address the challenges of transparency and instability in transferring to AI-supported communication.

One participant (P1) stated:

It is really ambiguous. I believe that if this type of AI technology can function effectively, the overall quality of nursing work will improve. AI can handle repetitive and time-consuming tasks, allowing nursing staff to focus more on complex tasks, such as managing emergencies, providing emotional support to residents, and ensuring their overall well-being. However, all the data may not be automatedly input by the machine but by the NH staff, one by one currently, which causes problems in providing residents’ real-time data to family members. We also are afraid that the need for time-transparent data may worsen our workload.

Four categories of manifest content emerged: ambiguity in replacing repetitive communication with AI (n = 15), ambiguity in emotional interaction (n = 13), ambiguity in transparent data sharing (n = 9), and ambiguity in tailoring personal care plans for families (n = 7). Each theme is presented below, along with representative quotes and interpretations.

Theme 1:

Ambiguity in replacing repetitive communication with AI.

Nurses frequently encounter repetitive inquiries from family members, particularly regarding residents’ health status and daily routines. These repeated nursing-family interactions disrupt nurses’ workflow and reduce efficiency, especially given the heavy workloads and multiple responsibilities they manage. Although nurses recognize the potential of AI-supported communication tools to reduce these redundancies, they also express concerns about whether current technologies are sufficient to meet families’ expectations for timely and detailed updates.

One participant (P1) stated:

There may be a gap between the expectations of family members and the services we can provide on a daily basis. Some family members expect us to pay attention to every small detail of the residents at all times. However, in reality, we have many residents to care for, and it's impossible to provide constant, detailed updates. Sometimes, family members repeatedly ask about the resident's dietary or sleeping conditions, even though these have been explained multiple times in previous communications. This affects our work efficiency, as we have to spend time on these repetitive inquiries instead of focusing on actual care. It's expected that digital tools could replace this type of work.

Another participant (P3) shared:

I might need to communicate with five to six family members per day, and many of their questions are repetitive. I understand their anxiety—it's natural for them to be concerned since they’re not here every day—but it does make my workload heavier. Especially when there's a change in a resident's condition, the number of inquiries increases significantly. It often feels like a group of family members is constantly asking similar questions. The challenge is that, sometimes, families are dissatisfied with how quickly we respond. They expect immediate updates, but with our heavy workload, it's not always possible to provide prompt responses. Lately, it feels like families view the healthcare industry as a service industry and treat us like customer service staff.

Although the participants held a positive attitude toward AI-supported communication tools that could potentially replace repetitive tasks, they also expressed concerns that the current technology requires improvement to reduce their workload. Most nurses reported that they did not use primarily electronic medical record systems or basic data monitoring tools. They also faced challenges such as cumbersome operations and data instability.

One participant (P1) stated:

The current AI systems can sometimes feel overly complicated, adding to my workload due to the need for manual data entry and organization. These systems are often cumbersome to operate, especially when technical issues arise, forcing us to manually record data, which ultimately increases our workload. Simplifying these processes would make AI much more practical and efficient in supporting our daily tasks. … Many of the current AI systems are still quite complex and require a significant amount of time to learn how to use them.

Theme 2:

Ambiguity in emotional interaction.

Managing the emotions of family members is often a significant challenge, as noted by the interviewees. Although the potential use of chatbots and other digital tools to enhance communication was acknowledged, nurses expressed a desire for these tools to provide personalized updates and emotional support to reassure anxious families—an aspect of their work that can be particularly stressful. This theme highlights nurses’ hope that AI-supported communication can assist with emotional engagement while also addressing their concerns about AI's ability to convey genuine empathy.

One participant (P5) stated:

If the chatbot could automatically notify family members about significant changes in the resident's condition and share updates or encouraging words, that would be amazing. This way, families wouldn’t feel that their loved ones are being neglected, and they would sense our genuine care for the residents. Such a feature could reduce our communication burden and improve family satisfaction. … I hope that, in the future, chatbots can become more humanized, showing greater empathy when interacting with family members. The chatbot could provide appropriate support and comfort based on the family's emotions or concerns, offering encouragement and advice tailored to the resident's specific situation. This would not only enhance the family's experience but also help us feel more supported during our communication efforts.

The participants also expressed uncertainty about whether AI could effectively replicate the empathy and emotional nuances required in family interactions. They hoped that AI-supported communication tools could use voice or text to comfort families. They also suggested that chatbots could help by automatically sending caring messages, updates, or reminders to reassure families about their loved ones’ well-being. As one participant (P9) stated:

I hope AI can improve in providing emotional support. While it cannot fully replace human emotional interaction, it could help by offering calming messages via voice or text to family members, especially when we cannot respond immediately. This would help reduce their anxiety. Family members often need care and reassurance. If AI could respond in a more empathetic way—like saying, “Your father's blood pressure is very stable today. We will continue to monitor his condition closely, so there's no need to worry too much”—it would transform a cold data transmission into a warm, comforting interaction.

Theme 3:

Ambiguity in transparent data sharing.

Participants expressed concern that sharing too much data—especially measurements recorded by nursing aides or information related to complex medical situations—could cause anxiety or confusion among family members. This could lead to more frequent inquiries and demands for clarification, thereby increasing the emotional and communication workload for staff. This theme highlights the delicate balance that nurses must strike between promoting transparency and preventing miscommunication or unnecessary distress for families.

One participant (P13) stated:

In institutional care settings, the workload for nurses is extremely high, and, sometimes, health measurements are not updated in real time. If families have access to this data at any moment, they might question why certain measurements, such as blood pressure or fluid intake, haven't been uploaded yet. Additionally, in many institutions, non-nursing staff often perform these measurements, raising concerns about their accuracy.

Nurses also emphasized that the accuracy and stability of AI technologies are critical. System malfunctions or the delivery of incorrect information can disrupt workflows and undermine trust in digital tools. These concerns highlight the need for reliable systems and clear backup protocols to effectively manage potential errors. Without careful management, efforts to increase data transparency through digital tools could paradoxically increase, rather than reduce, nurses’ communication workload.

As another participant (P15) stated:

Moreover, some families lack a full understanding of the clinical context. If they see even minor changes or alarming indicators, such as red flags in the data, they may become excessively anxious and rush to inquire, creating stress for the nursing staff. … Families may not understand these nuances and, after researching online, might repeatedly call nurses with concerns over minor variations. … The added pressure of frequent, unnecessary inquiries could overwhelm caregivers, potentially impacting their ability to focus on critical care tasks.

Theme 4:

Ambiguity in tailoring residents’ care plan for families.

Participants expressed uncertainty about using conversational chatbots to tailor residents’ care plans for families, reflecting concerns about balancing automation with the need for professional judgment. Although they were hopeful that chatbots could automatically generate detailed reports from residents’ health data, offer practical advice, and deliver personalized health suggestions or alerts, they questioned whether such tools could adequately replace clinical expertise. This theme highlights nurses’ interest in the predictive and anticipatory capabilities of AI, tempered by concerns about its limitations in capturing the nuances of individualized care.

As one participant (P6) stated:

What I hope for in a chatbot are innovative features, such as automatically generating detailed reports based on a patient's health data and providing practical advice. The chatbot could offer personalized health suggestions or alerts derived from medical records and health conditions, giving families access to more useful information. AI systems should have stronger data analysis capabilities, such as the ability to identify potential health issues from large volumes of resident health data and alert us in time to take appropriate action. The chatbot could also analyze a family's inquiry history and proactively prepare relevant information even before questions are asked. This predictive, anticipatory feature would significantly reduce our manual communication workload.

Another participant stated (P9):

Oh, I think a generative health management chatbot could be really helpful. I believe AI technology should be developed toward being smarter and more humanized. AI systems should have stronger data analysis capabilities, enabling them to identify potential health issues from large amounts of resident health data and provide timely alerts for intervention. This would alleviate our workload by helping us prevent problems rather than waiting to address them after they occur. … This would eliminate the need for manual tracking and adjustments, saving us valuable time. Especially, it could handle many common questions, and since the chatbot can work 24-7, families would be able to get the answers they need at any time, without having to wait for our working hours to respond. I believe such innovations would make the chatbot a more effective communication tool.

Further, participants expressed concern that AI might miss subtle emotional or behavioral changes in residents that require human observation and judgment, potentially overlooking early signs of health deterioration. They feared that over-reliance on AI could reduce their active engagement with families, leading to a loss of emotional connection and weakening their professional role in caregiving. These concerns illustrate nurses’ nuanced view that, although AI can support the development of care plans, human oversight is essential to ensure safe, empathetic, and personalized care. As one participant (P3) noted:

I do have some concerns as well. If we become overly reliant on AI systems, we might lose the importance of ourselves in communication with family. Further, AI might not detect subtle emotional changes in residents, which could be early signs of a decline in their health. Therefore, while adopting AI systems, we still need to maintain a high level of vigilance and professional judgment, ensuring that technology doesn’t entirely replace our role.

Discussion

This study identified a central theme: navigating ambiguities in role transformation, which highlights the dual perspectives of nurses toward AI-supported communication tools in NHs. Nurses recognized that AI could help to streamline routine communication tasks and improve access to residents’ information. They also expressed concern, however, about AI's ability to preserve emotional connection, clarify professional responsibilities, and support personalized interactions with families.

Ambiguity can arise in the context of the multifaceted and complex roles that nurses fulfill in NHs. Communication with family members extends beyond the exchange of clinical information; it also involves addressing emotional needs, cultural expectations, and family dynamics. Nurses must navigate diverse perspectives on caregiving and cultural attitudes toward aging and end-of-life issues, as well as varying levels of trust and involvement among family members. As a result, effective communication in NHs requires not only clinical expertise but also relational sensitivity, cultural competence, and ethical decision making.

These findings align with the unified theory of acceptance and use of technology, which posits that users’ acceptance of technology is influenced by four constructs: performance expectancy, effort expectancy, social influence, and facilitating conditions.³³ Although nurses recognized the potential of AI-supported tools to improve communication efficiency (performance expectancy), their concerns highlight barriers related to effort expectancy, particularly when engaging in emotionally sensitive or ethically complex interactions. Nurses may perceive AI systems as inadequate if they fail to support the relational, empathetic, and context-sensitive aspects of communication with families, which are fundamental to their caregiving roles.

The first theme reflected that nurses were ambivalent about replacing repetitive communication with digital tools. One key reason is that families often place residents in NHs primarily for health monitoring and care.³ Consequently, they expect frequent and detailed updates on residents’ conditions, leading to repetitive communication between family members and nursing staff. Although digital tools have the potential to reduce this communication burden, nurses noted several limitations in their current application, including poor system integration, inadequate technical infrastructure, and workflows that do not align with the daily realities of care delivery.

Although EHRs are often promoted as a solution to improve care quality and communication, their adoption in NHs has been slow.³⁴ For instance, <70% of NHs in the United States have implemented EHR systems, and about 5% of these facilities have no plans to adopt them.³⁵ Adoption is more common in better-resourced facilities or those that face competitive pressures. Even in settings in which EHRs are in place, the increase in technology use during the pandemic has not always improved staff readiness or confidence in using these tools for family communication.

To address these challenges, policymakers and NH administrators should consider a phased and context-sensitive approach to implementing AI-supported communication tools. The use of semi-automated, usability updates tailored to the needs of long-term care environments may ease adoption. Successful implementation will require adequate staff training, clearly defined roles, and technology systems that integrate smoothly into existing care workflows. Moreover, a deeper understanding of the technological maturity across different types of facilities can guide the design and evaluation of appropriate solutions. Importantly, investments in digital infrastructure—particularly in rural or under-resourced NHs—are essential to ensure equitable access to effective EHR systems and communication technologies.

The results also revealed that ambiguity in emotional interaction was a concern. Recent research, however, suggests that chatbots have significant potential to provide social and psychological support in real-world human interactions, but their effectiveness depends on the level of service they are designed to offer.³⁶ Although natural language processing tools, such as ChatGPT and other AI-generated content, can be useful in certain healthcare contexts,5 they should not replace human healthcare providers entirely and must be used with caution in such settings. This highlights the need for further research into the capacity for emotional communication in digital tools. Although automation can enhance efficiency, these systems must be designed with emotional sensitivity, particularly when families seek reassurance or empathy from healthcare professionals.

Our results also showed that nurses experienced ambiguity in regard to transparent data sharing. Transparent and detailed communication from the staff is critical for families.^5,6 This ambiguity underscores the need for communication systems that allow selective and contextualized data sharing to address informational needs and emotional sensitivities effectively. Excessively detailed health information, if not appropriately contextualized, may cause family members undue anxiety and misinterpretation, increasing communication burdens on staff. This misunderstanding often results in frequent inquiries and increases the emotional burden on the nursing staff.

In particular, communication issues between nursing staff and families sometimes involve conflicting views—differences in the viewpoint of NH services, nurturing health, and medication between family members and others10—and the extent of data revealed to the family needs to be carefully considered. Striking a balance between delivering essential health information and addressing the emotional sensitivities of families remains a critical challenge in AI-supported communication. These findings highlight the importance of designing data transparency strategies that are supported by empirical research, ensuring that shared information is accessible and reassuring for families, while alleviating the strain on caregivers. The level of data transparency requires an accurate assessment to provide a baseline understanding for communication planning.

Ambiguity in tailoring residents’ care plans for families was a key concern of participants. This challenge may stem from nurses’ unfamiliarity with the integration of AI into their work processes. The use of tools such as ChatGPT in healthcare settings has shown potential to improve patient care and satisfaction by providing personalized, accurate, and up-to-date information.^5,37 As such, AI-related education is essential for nurses, enabling them to use AI as an assistive tool in their roles and address AI-supported communication challenges, such as handling issues with chatbots and ensuring accurate data input into the system. Specifically, structured training programs for NH staff should be developed to enhance their digital skills, allowing them to effectively supervise AI-generated content and ensure safety and person-centered care planning.

Limitations

Although this study offers valuable insight into nurses’ expectations regarding AI-supported communication with families in NHs, several limitations should be acknowledged. First, due to the qualitative research design, no statistical sample size calculation or power analysis was conducted. While this approach is consistent with qualitative methodology—where the focus is on depth and thematic saturation rather than numerical representation—it may limit the generalizability of the findings. However, the findings may support case-to-case transferability. By providing detailed contextual descriptions of the research setting, participant characteristics, and data collection process, we enable readers to assess the relevance and applicability of the findings to similar long-term care contexts or populations.

Second, all participating facilities were medium-to-large NHs located in rural areas of Taiwan, reflecting a common model of long-term care in non-urban Taiwan, where such facilities are more common due to lower land and operating costs. This rural focus may have influenced nurses’ perceptions, as rural settings often differ from urban ones in terms of digital infrastructure, staffing levels, and the frequency of family visits. Third, none of the participating NHs had fully automated EHR systems in these facilities, which may have introduced bias in nurses’ expectations regarding digital communication. Nurses working in such settings may have had prior exposure to partial or fragmented digital systems, which could influence their perceptions—either positively, due to perceived potential, or negatively, due to past limitations—thus affecting the generalizability of the findings to other long-term care contexts. Finally, the findings of this study are shaped by Taiwan's cultural context, particularly the strong emphasis on family involvement in care decisions rooted in the value of filial piety. This cultural expectation may influence how nurses perceive their roles and responsibilities in AI-supported communication with residents’ families. Consequently, concerns such as emotional disconnection, unclear role boundaries, and the risk of depersonalized care may emerge differently in healthcare systems with varying cultural norms and communication practices. Future research should explore how cultural and organizational factors influence the design, implementation, and acceptance of AI-supported communication tools in long-term care settings. Including diverse facility types and geographic regions, such as urban environments, is essential to ensure the broader applicability of the findings.

Conclusion

This study explored nurses’ expectations of AI-supported communication with residents’ families in NHs. The core theme, navigating ambiguities in role transformation, captured their mixed attitudes: Although nurses saw the potential of AI to reduce repetitive communication and improve access to information, they also voiced concerns about emotional disconnect, unclear responsibilities, and the risk of depersonalized care. To ensure successful implementation, AI tools should be designed to support, not replace, nurses’ communication roles. Addressing these ambiguities is key to developing digital systems that reflect the realities of care work. Based on the findings, we recommend a phased approach to digital adoption that begins with usability, semi-automated tools that integrate with existing workflows. Adequate staff training, clear role definitions, and investment in digital infrastructure, particularly in under-resourced or rural settings, also are essential. These strategies can help to bridge the gap between technological promise and frontline usability, ensuring that AI tools enhance care efficiency and human connection.

Footnotes

Acknowledgements

We would like to thank all participants and the selected nursing homes for supporting this study.

ORCID iDs

Yu-Yi Lin

Hsiu-Hsin Tsai

Wann-Yun Shieh

Li-Chueh Weng

Hsiu-Li Huang

Chia-Yih Liu

Ethical considerations

This study was approved by the Chang Gung Medical Foundation institutional review board (approval no.202400678B0) on 13 May 2024.

Consent to participate

All participants provided informed written consent prior to participating.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the National Science and Technology Council (NSTC 113-2314-B-182-035-MY3) and the Chang Gung Memorial Hospital (BMRP849).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The data are not publicly available due to privacy or ethical restrictions.

References

Mitchell

Kemp

. Quality of life in assisted living homes: a multidimensional analysis. J Gerontol B Psychol Sci Soc Sci 2000; 55: 117–127.

Puurveen

Baumbusch

Gandhi

. From family involvement to family inclusion in nursing home settings: a critical interpretive synthesis. J Fam Nurs 2018; 24: 60–85.

Tsai

. Family members’ perceived meaning of visiting nursing home residents in Taiwan. J Adv Nurs 2012; 68: 302–311.

Ryan

McKenna

. ‘It's the little things that count’. Families’ experience of roles, relationships and quality of care in rural nursing homes. Int J Older People Nurs 2015; 10: 38–47.

Gaugler

Mitchell

. Reimagining family involvement in residential long-term care. J Am Med Dir Assoc 2022; 23: 235–240.

Roberts

Ishler

. Family involvement in the nursing home and perceived resident quality of life. Gerontologist 2018; 58: 1033–1043.

Haesler

Bauer

Nay

. Staff-family relationships in the care of older people: a report on a systematic review. Res Nurs Health 2007; 30: 385–398.

Patel

Landrigan

. Communication at transitions of care. Pediatr Clin N Am 2019; 66: 751–773.

Bowers

Chu

, et al. What COVID-19 innovations can teach us about improving quality of life in long-term care. J Am Med Dir Assoc 2021; 22: 929–932.

10.

Tsai

Weng

, et al. More than communication skills: experiences of communication conflict in nursing home nurses. Med Educ 2013; 47: 990–1000.

11.

Tsai

Cheng

Weng

, et al. Motivations for family members’ visits to nursing home residents with dementia in Taiwan: a qualitative study. J Nurs Scholarsh 2021; 53: 87–95.

12.

Schnelle

Bates-Jensen

Levy-Storms

, et al.

The minimum data set prevalence of restraint quality indicator: does it reflect differences in care?

Gerontologist 2004; 44: 245–255.

13.

Bao

Sun

Meng

, et al. 2019-nCoV epidemic: address mental health care to empower society. Lancet 2020; 395: e37–e38.

14.

Turale

Nantsupawat

. Clinician mental health, nursing shortages and the COVID-19 pandemic: crises within crises. Int Nurs Rev 2021; 68: 12–14.

15.

Ornell

Schuch

Sordi

, et al. “Pandemic fear” and COVID-19: mental health burden and strategies. Braz J Psychiatry 2020; 42: 232–235.

16.

Said

El-Shafei

. Occupational stress, job satisfaction, and intent to leave: nurses working on front lines during COVID-19 pandemic in Zagazig City, Egypt. Environ Sci Pollut Res Int 2021; 28: 8791–8801.

17.

Munyisia

Hailey

. How nursing staff spend their time on activities in a nursing home: an observational study. J Adv Nurs 2011; 67: 1908–1917.

18.

Tsai

Huang

. Nursing home nurses’ experiences of resident transfers to the emergency department: no empathy for our work environment difficulties. J Clin Nurs 2016; 25: 610–618.

19.

Vermeir

Degroote

Vandijck

, et al. Job satisfaction in relation to communication in health care among nurses: A narrative review and practical recommendations. Sage Open 2017; 7.

20.

Srinivas

Chakkravarthy

Battacharyya

. Artificial intelligence approaches for the COVID-19 pandemic. In: Garg

Chakraborty

Mahmoudi

Sohmen

(eds) Healthcare informatics for fighting COVID-19 and future epidemics. Cham: Springer Publications, 2022, pp.249–267.

21.

Hugelius

Harada

Marutani

. Consequences of visiting restrictions during the COVID-19 pandemic: an integrative review. Int J Nurs Stud 2021; 121: 104000.

22.

Kemp

. Report: Most important data on digital audiences during coronavirus. Growth quarters. https://thenextweb.com/growth-quarters/2020/04/24/report-most-important-data-on-digital-audiences-during-coronavirus/ (2020, accessed 01/12/2025).

23.

Koeze

Popper

. The virus changed the way we Internet. The New York Times. https://www.nytimes.com/interactive/2020/04/07/technology/coronavirus-internet-use.html (2020, April 7, accessed 01/10/2025). 2020, April 7.

24.

Santandreu-Calonge

Medina-Aguerrebere

Hultberg

, et al.

Can ChatGPT improve communication in hospitals?

El Profesional de la Información 2023; 32.

25.

Ministry of Health and Welfare. Number of general nursing home institutions. https://nurse.mohw.gov.tw/cp-179-1123-085a1-2.html (2025, accessed 01/10/2025).

26.

Taiwan Long-term Care Professional Association. https://www.ltcpa.org.tw/ (accessed 04/05/2025)

27.

Sandelowski

. Sample size in qualitative research. Res Nurs Health 1995; 18: 179–183.

28.

Holloway

Wheeler

. Qualitative research in nursing. 2nd ed. Oxford: Blackwell Publishing, 2002.

29.

Patton

. Qualitative Research and Evaluation Methods. 3rd ed. Thousand Oaks, CA: Sage Publications, 2002.

30.

Graneheim

Lundman

. Qualitative content analysis in nursing research: concepts, procedures and measures to achieve trustworthiness. Nurse Educ Today 2004; 24: 105–112.

31.

Elo

Kääriäinen

Kanste

, et al. Qualitative content analysis: a focus on trustworthiness. SAGE Open 2014; 4: 1–10.

32.

Lincoln

Guba

. Naturalistic Inquiry. Thousand Oaks, CA: Sage Publication, 1985.

33.

Venkatesh

Morris

Davis

, et al. User acceptance of information technology: toward a unified view. MIS Q 2003; 27: 425–478.

34.

Pradhan

Dayama

Morris

, et al. Enhancing nursing home quality through electronic health record implementation. Health Inf Manag 2024: [Online ahead of print].

35.

Felix

Dayama

Morris

, et al. Organizational characteristics and the adoption of electronic health records among nursing homes in one southern state. J Appl Gerontol 2021; 40: 481–488.

36.

Haque

MDR

Rubya

. An overview of chatbot-based Mobile mental health apps: insights from app description and user reviews. JMIR Mhealth Uhealth 2023; 11: e44838.

37.

Haleem

Javaid

Singh

. An era of ChatGPT as a significant futuristic support tool: a study on features, abilities, and challenges. Benchcouncil Trans Benchmarks, Standards and Evaluations 2022; 2: 100089.