It is critical to understand patients’ perceptions, as they directly assess the risks, benefits, and barriers involved in using these AI tools. In response, we propose a hypothetical research framework, grounded in existing literature, to explore the factors that may affect the motivations and intentions of adopting AI-based home care systems. This framework incorporates 5 constructs: privacy, accountability and security, attitude, perceived usefulness and comfortability, and motivation to adopt to fill the research gap and inform stakeholders of consumers’ needs and concerns.

AI-based home care systems collect and process real-time personal health data, facilitating human-computer interactions and patient health monitoring [13]. However, privacy concerns arise since users are understandably sensitive to personal data [11,14]. Privacy considerations revolve around how information is collected, stored, accessed, and shared [1,4]. These concerns could discourage individuals from sharing information and using health services, thereby hindering the widespread adoption of AI in health care delivery [15]. Beyond technology, addressing patients’ rights to oversee their data in our increasingly digital world is imperative. Crucially, regulatory compliance is situated under the umbrella of privacy because it is a crucial mechanism that enforces adherence to established data protection standards. Regulatory mandates, often developed in response to public concerns about data privacy, work to ensure that personal data are well handled [14]. Regulatory compliance is not just about legal obedience; it gives individuals a sense of assurance that their data are being managed with integrity and transparency. This underscores the pressing need for stringent regulations governing patient data acquisition, processing, and storage [16,17]. The degree of regulatory compliance and level of privacy anxiety may impact the perceived comfortability and attitude toward AI adoption. As such, our study considers 3 dimensions of privacy issues: perceived comfortability with information storage, data collection practices, and perceived regulatory compliance.

Despite the increasing prevalence of research on AI governance issues, there is a lack of studies considering patients’ perceptions of accountability issues in this context. The lack of clear accountability for the actions of AI may create a sense of insecurity and unease for patients [18-20]. While the ongoing dialog on AI governance is becoming increasingly pertinent, there remains a notable gap in comprehending patients’ perspectives, particularly regarding the accountability and security of AI applications. The confluence of accountability and security is intentional. Accountability revolves around the notion of answerability—determining who or what entity bears the onus when AI decisions go awry. Security, on the other hand, focuses on safeguarding patient information from unwarranted access or breaches. These 2 facets are intertwined; without a transparent system of accountability, the integrity of data security is compromised. For instance, if an AI system makes a decision leading to a patient’s harm and there is no clear entity to hold accountable, it implies potential lapses in data security and the AI’s operational parameters. Navigating these complexities poses significant challenges. A lack of consensus solutions exacerbates patients’ fears about data misuse and the trustworthiness of AI systems [20]. Moreover, the inherent complexity of AI, which often results in opaque validation processes, may magnify these concerns [1,21]. Additionally, unlike humans, AI lacks subjective consciousness in its decision-making. This absence positions AI as a tool rather than an active participant with intent. Consequently, questions arise about the responsibility and accountability for AI-driven decisions, creating patient concerns about the security and reliability of relying on AI [19]. Therefore, our study explores patients’ perspectives on these concerns and examines 4 dimensions of accountability and security: data security and use, patients’ rights regarding their medical records, AI developer accountability, and physician or hospital accountability.

Perceived usefulness, a core construct of the technology acceptance model, is crucial in evaluating technology acceptance [22,23]. In addition to perceived usefulness, this study introduces comfortability as a significant factor. We define comfortability as the degree to which patients perceive the AI-based home care systems to be comfortable for managing chronic conditions and promoting personal health status [24]. We hypothesize that patients are more likely to adopt a technology when they perceive it as beneficial (usefulness) and feel at ease and secure while using it (comfortability). In this study, these can be expressed as the degree to which the patients perceive the AI-based home care systems are useful and comfortable for managing chronic conditions and promoting personal health status [24]. Consumers evaluate usefulness based on perceived benefits and convenience [12,25] and expect enhanced communication with physicians when AI provides more information about their health status [1]. Additionally, patients expect cost reduction in long-term care while maintaining recovery quality with AI-based home care systems [26]. Furthermore, AI systems offer unlimited access to technical education and health knowledge, providing positive guidance and enhancing overall patient comfort and usefulness [23]. This measure contributes to the proposed model by capturing patients’ perception of the system from these 4 aspects: reducing health care costs, facilitating understanding of health conditions, improving communication with care providers, and educating patients about their health.

Successful adoption of AI-based home care systems requires an examination of patients’ attitudes and perceptions of AI [27,28]. Attitudes, which are deeply entwined with patients’ perceptions of the technology, directly influence their intention to use and motivation to accept these systems [23]. Trust is critical to patients’ attitudes toward AI, particularly when considering the balance between safeguarding personal information and receiving personalized services and treatment [29]. Moreover, patients’ comfort level with AI’s role in their treatment and their daily use frequency are also crucial in determining their attitude toward AI. If patients feel comfortable receiving the medical results from AI participated diagnoses, especially for serious diseases, a positive attitude may be fostered to alleviate doubts and distrust of the adoption. This study incorporates the attitude construct in the proposed model by examining it from 4 perspectives: attitude of daily use, attitude of AI’s future role, attitude of trust, and attitude of receiving serious diagnoses from AI.

In summary, we incorporate constructs drawn from the existing literature and studies, comprising 5 main constructs: privacy, accountability and security, attitude, perceived usefulness and comfortability, and motivation to adopt.

The following hypotheses are proposed to explore the key relationships between these constructs:

Table 1 outlines respondents’ demographic characteristics in detail. The data show a relatively balanced gender distribution, with 52.3% (n=116) males and 47.7% (n=106) females, respectively. Over half of the respondents fall within the 31-45 years age group, suggesting a concerning trend of chronic illnesses among younger individuals. The respondents’ racial composition aligns with the US Census Bureau’s report from July 2021; for instance, the percentages of self-identified White Americans from the respondents and the Census Bureau are around 72.5% and 75.8%, respectively [41]. Around 80% (n=176) of the respondents in our survey have achieved at least a bachelor’s degree, which might be indicative of a selection bias, given that MTurk platform users tend to be more educated than the average working adult population [30]. In terms of income, the majority of the respondents fall into the ranges of US $25,000-US $50,000 (n=76, 34.2%) and US $50,000-US $100,000 (n=89, 40.1%), aligning with the US median household income [42].

The results supported 5 of 7 research hypotheses. The perceived usefulness and comfortability of AI-based home care systems had a direct, significant impact on patients’ motivation to adopt AI-based home care systems and an indirect influence through altering their attitudes toward AI. Besides, we observed that concerns about privacy and accountability issues may influence patients’ motivation to adopt through the usefulness and attitude toward adoption, which aligns with the previous findings [15,23]. Consumers’ privacy concerns highly impacted the perceived usefulness and comfortability (P<.001), corroborating earlier studies [15,33]. If AI systems were designed with adequate security and regulated to respect patients’ privacy, they perceived the system as more comfortable and usable.

Interestingly, privacy issues did not significantly affect consumers’ attitudes toward using AI-based home care systems (P=.21). One possible explanation could be that the direct relationship between privacy and attitude is overshadowed by other influential factors, such as perceived comfortability and perceived usefulness. The novelty of AI-based home care technology might be captivating users’ attention, causing them to prioritize its perceived benefits over potential risks. Furthermore, consumers are often known to trade off privacy for convenience, especially when the potential risks are not immediate or tangible. Given that interactions with AI are often more intuitive than the abstract concept of privacy, consumers may overlook privacy concerns until a data breach or misuse occurs [7,34]. At this stage, the perceived usefulness of AI-based home care systems temporarily outweighs privacy concerns. Additionally, the perception of privacy has been evolving rapidly in the digital age, with many consumers desensitized to data collection practices.

The issue of AI accountability is also a controversial issue in health care, as it is unclear who should hold responsibility for AI’s actions [35]. This study showed that accountability issues directly influence patients’ attitudes toward using AI-based home care systems (P<.001), adding unique insights to the current literature. Patients who were highly concerned about the responsibility issue tended to develop a more negative attitude toward using AI-based home systems. This suggests that clear regulations around responsibility would be enacted to enhance the usage confidence [15], which is supported by the early findings related to technology adoption in health care [49,50]. However, we did not find a significant effect of accountability on perceived usefulness and comfortability (P=.10). One possible explanation is that while accountability is crucial for trust-building, its impact is perhaps more indirect in nature. Patients may conceptualize accountability as a macro-level concern, pertinent mainly to regulators and AI developers. Thus, it may not directly translate to their perceptions about how useful or comfortable an AI system is for their day-to-day needs. This suggests that even though patients desire a clear understanding of who is accountable during system errors, they may not see these concerns as directly affecting the immediate advantages or their perception of the utility and comfort of AI-based home care systems. Moreover, it is possible that patients assume that once the technology has been approved and is available on the market, the accountability issues have been duly addressed by relevant authorities [19]. Hence, while accountability concerns can affect their general attitude, it does not seem to permeate their evaluation of the system’s practicality or convenience. For a comprehensive embrace of AI systems in home care, it is paramount that governance bodies understand these nuanced reactions to accountability, recognizing that a perceived lack of it could impair patient trust [35].

On the other hand, patients with chronic diseases desire AI to offer convenience and usefulness in health management at home rather than going to clinics with long waiting times [26]. Consistent with prior research [25,51], this study reaffirmed that the motivation of patients to adopt AI-based home care systems stems from the perceived usefulness and comfortability of these systems (P=.007) as well as the attitudes toward the adoption (P=.002). Furthermore, we also concluded that perceived usefulness and comfortability was strongly associated with the performance expectancy on attitude (P=.003), consistent with the previous study [23]. Thus, for potential consumers with chronic diseases, recognizing the practicality of AI-related systems fosters positive attitudes toward acceptance, enhancing adoption motivation [31,52].

As the developers and distributors of AI-based home care systems, care or technology providers have much earlier access to the system than the end-user patients. It has always been a challenge to develop AI-based home care systems that meet the majority of end users’ expectations. However, they can still proactively anticipate and address user needs, which is crucial in facilitating user adoption and satisfaction. In this context, this study offers valuable implications.

While it is widely acknowledged that any novel technology should provide comfort and use, this study suggests that user’s trust in the systems’ functionality and ethical integrity can also positively impact adoption decisions [53]. The care or technology providers are responsible for developing a reliable, interpretable system to alleviate user anxiety. Since the entire AI process is similar to a black box, care or technology providers should work to validate the AI algorithms and present them more understandably if needed [10]. This implies that care or technology providers should design and implement secure data storage and transmission mechanisms, making it transparent and clear for users how their data are used and protected. Care or technology providers should also empower users with control over their own data, allowing them to view, correct, and delete their data as needed [40].

Importantly, the primary role of AI at this stage is not to replace but to supplement and enhance primary care. The design of AI systems should be patient-centric, taking into account the diverse needs of individuals with chronic conditions. A system customizable to various health conditions, lifestyles, and user preferences can foster a sense of personalization and thus promote engagement and long-term use [53]. By providing tools with clear, concise, and user-friendly instructions, AI can guide patients to improve doctor-patient communication and make care delivery more cost-effective, resulting in efficient doctor-AI-patient interactions.

Moreover, comprehensive and straightforward education and ongoing support should be personalized based on the individual user’s health condition and learning capability [54]. It is important that patients understand their role and have the necessary information to make informed choices rather than being passive AI recipients. Guidelines in this regard can increase patient interest in AI use and their adoption intentions. The regular feedback from patients is also crucial for continuous improvement. Providers can leverage AI technologies to capture real-time user feedback and use these data to refine the system continuously.

AI’s emergence in health care has not been met with timely policy adaptations, as technology often outpaces regulatory responses [1,19]. This study has investigated patients’ perceptions of the regulation and governance to provide insights to policy makers for better adaptation in AI-based home care.

One of the biggest concerns patients have is about the management of their medical data by AI-based home care systems. Concerns primarily revolve around data sharing, exchange, and their ethical implications. These emerging issues challenge traditional health care ethics, requiring policy makers to balance the potential benefits against patients’ privacy rights. To address these challenges, policy makers are advised to clearly define the legal and ethical boundaries of data collection, storage, use, and sharing. Establishing and enforcing standards and certification mechanisms for AI systems’ safety, effectiveness, and compliance would be prudent. Policy makers must ensure that patients are fully informed about the data that are being collected, why it is being collected, and how it will be used, and that they can make informed decisions when using AI-based home care systems.

Moreover, accountability in the current governance system is unclear, particularly in defining AI involvement in decision-making for care delivery and the extent of responsibility for biases and errors. Any unclear and opaque responsibility delineation could undermine patients’ trust and further impact perceived comfortability [19]. A clear accountability guideline should address issues such as who is responsible for the AI recommendation errors and how to handle bias results in unfair treatment or outcomes for certain groups of patients. In such contexts, while AI developers must uphold and strive for the highest precision standards, the primary accountability for the decision-making process would logically reside with the health care professionals. On the other hand, in situations where AI systems are designed to play a more independent role, particularly in remote patient monitoring setups without immediate human oversight, the responsibility might predominantly fall on the AI providers because their systems function autonomously without human checks. Establishing clear guidelines in these areas would likely enhance patients’ trust in and willingness to adopt AI solutions.

This study has a few limitations. First, this study used a sample from a crowdsourcing marketplace in the United States. There is a challenge in verifying the authenticity of the health conditions claimed by respondents. Moreover, using MTurk may have introduced a certain degree of sample bias, limiting the generalizability of our findings. To ensure the accuracy of our data, we initially sampled over 300 individuals, though we acknowledge the inherent limitations in fully verifying the chronic condition status of respondents. Specifically, a considerable proportion of our respondents were relatively young, ranging from 31 to 45 years old, and were well-educated, with approximately 80% (n=176) possessing bachelor’s degrees or advanced degrees. This demographic distribution may not represent the typical profile of patients with chronic diseases, who are often older and display a broader range of education levels [16,17]. Such discrepancy highlights the potential anomaly in our sampling strategy and suggests caution in interpreting results with broader, more diverse populations. Furthermore, some patients may experience multiple chronic conditions simultaneously. This complexity could have significant implications on the required health care resources and the patients’ attitudes toward AI-based home care systems. Future studies could aim to understand patients’ diverse health conditions and varied health care demands to deepen our understanding of patients’ acceptance of AI-based home care systems. However, this limitation does not detract from the significance and originality of this work within the scope of the defined sample.

Moreover, in future research, we plan to incorporate more rigorous verification mechanisms, such as requiring medical documentation or collaborating with health care institutions, to ensure the authenticity of participants’ health conditions. This will provide a more robust data collection foundation and further strengthen our research outcomes’ validity. Future research could also aim to explore more diverse and representative patient samples, considering the variations in backgrounds and health care demands.

AI-based home care systems are a promising development in health care, potentially improving the delivery and accessibility of care for patients with chronic diseases. Our findings indicate that patients have an overall positive perception of AI-based home care systems, and their motivation to adopt such systems is significantly influenced by the perceived usefulness and comfortability and their attitude toward use. However, persistent concerns around privacy and accountability underscore the need for improved data management and comprehensive regulations. This study provides invaluable insights for a range of stakeholders, including policy makers, health care providers, and patients, to effectively and ethically use AI-based home care systems. As the field evolves, research should continue to refine and expand upon these insights, enabling us to leverage AI’s potential to enhance health care outcomes fully.