Sage Journals: Discover world-class research

Abstract

In the face of a rapidly aging population and the increasing demand for elderly care, the adoption of artificial intelligence (AI) in healthcare products has emerged as a promising solution to enhance service delivery. This paper investigates the behavioral evolution of multiple stakeholders—namely, government entities, AI healthcare enterprises, and medical professionals—in the adoption process of AI-enabled elderly care products. By employing an evolutionary game theory model, we analyze the stability strategies of these stakeholders under varying initial conditions. Our findings reveal that government subsidies and regulatory measures play a crucial role in promoting the adoption of these technologies, while the attitudes of enterprises and medical professionals are significantly influenced by perceived costs and benefits. Simulation analyses were conducted using MATLAB 2019a to validate the model, providing insights into optimizing stakeholder engagement and enhancing the adoption of AI in elderly care. We propose actionable recommendations for policymakers and industry leaders to foster the integration of AI into elderly care services, addressing critical challenges and leveraging opportunities in this evolving landscape.

Keywords

evolutionary game AI medical and health care product government subsidies enterprises’ attitude doctors’ adoption

What do we already know about this topic?

Based on the perspective of the dynamic evolutionary game, in this paper, a model is constructed for simulating the stability strategies of 3 subjects (the government, AI medical and health care enterprises, and the doctors) under different initial willingness levels.

How does your research contribute to the field?

In the development of artificial intelligence (AI) information technologies context, promoting the wide adoption of AI medical and health care products is posing an urgent problem.

What are your research’s implications towards theory, practice, or policy?

Countermeasures and suggestions are proposed on how to effectively strengthen the role AI medical and health care products and services play in solving the aging problem from the perspectives of the government, enterprises, and doctors.

Introduction

The rapid global demographic shift towards an aging population has intensified the need for innovative solutions in elderly care. According to the United Nations’ “World Population Prospects: The 2021 Revision,” the global population aged 65 and above is projected to surpass 1.5 billion by 2025, representing 16% of the total population. This demographic trend presents significant challenges, particularly in the areas of healthcare and social support systems. In China, these challenges are exacerbated by a declining fertility rate and a rapidly aging population, putting immense pressure on the nation’s pension and healthcare infrastructures.

In response to these challenges, the integration of artificial intelligence (AI) into healthcare, particularly in the development of AI-enabled elderly care products, has gained significant attention. These technologies offer the potential to enhance the quality of life for the elderly by providing more efficient and personalized care. The Chinese government, recognizing the importance of this integration, has implemented the “Action Plan for the Development of the Smart Health and Elderly Care Industry (2021-2025).” This plan emphasizes the development of an innovative industrial ecosystem that leverages AI and other advanced technologies to address the pressing needs of elderly care.Despite the potential benefits, the adoption of AI in elderly care is not without challenges. High development costs, concerns about data security and privacy, and the complexities associated with clinical efficacy pose significant barriers. Additionally, the adoption process is influenced by the behaviors and interactions of multiple stakeholders, including government entities, AI enterprises, and medical professionals.^1
-3 While previous studies have explored the role of AI in healthcare, there is a notable gap in the literature regarding the dynamic interactions among these stakeholders and how these interactions influence the adoption of AI-enabled elderly care products.

This study seeks to address this gap by developing a comprehensive understanding of the behavioral evolution of these stakeholders within the context of AI adoption in elderly care. The study employs an evolutionary game theory approach to model the strategic interactions among government entities, AI healthcare enterprises, and medical professionals. By analyzing these interactions, the study aims to uncover the factors that significantly impact the willingness of medical professionals to adopt AI technologies, thereby offering insights into the conditions necessary for widespread adoption.The timeliness of this research is underscored by the ongoing global push towards digital innovation in response to challenges such as the aging population and the increasing prevalence of chronic diseases among the elderly. The COVID-19 pandemic has further accelerated the adoption of digital technologies in healthcare, highlighting the critical need for robust and scalable solutions that can withstand future public health crises. This study differentiates itself from previous research by focusing on the evolutionary dynamics of stakeholder behavior rather than solely on the technological aspects of AI adoption. By doing so, it provides a more holistic view of the factors influencing the successful integration of AI into elderly care.

Related Literature

Opportunities and Benefits for Older Adults

Recent research has increasingly highlighted the benefits of AI and telehealth for older adults. These benefits include improved access to healthcare, enhanced management of chronic conditions, increased convenience, and overall better health outcomes. As the adoption of these technologies continues to grow, it is crucial to understand the specific advantages they offer to older populations, particularly in the context of the COVID-19 pandemic, which has accelerated the need for remote healthcare solutions.Its main advantages are:

Increased accessibility and convenience

Cheung et al⁴ examined older adults’ perceptions of telehealth services and found that such services significantly enhance accessibility to healthcare by reducing the need for physical travel. This study aligns with Peek et al,⁵ who identified factors influencing the acceptance of aging-in-place technologies, highlighting the importance of convenience in technology adoption. Greenhalgh et al⁶ provided insights into the rapid shift to video consultations during the COVID-19 pandemic, further supporting the role of telehealth in maintaining healthcare access amidst restrictions.

Enhanced health monitoring and management

Dicianno et al⁷ emphasized the benefits of mobile health technologies and smart home applications in continuous health monitoring and personalized health management. Their findings are supported by Husnain et al,⁸ who discussed the predictive capabilities of wearable technologies in monitoring medical events. Bonoto et al⁹ reinforced the effectiveness of mobile health interventions in managing chronic conditions such as diabetes and promoting physical activity, respectively.

Improved independence and quality of life

Kim and Lee¹⁰ explored the role of smart devices in managing chronic diseases, finding that these devices contribute to improved independence and quality of life among older adults. Sintonen and Immonen¹¹ identified critical factors influencing the adoption of telecare services, which enhance older adults’ ability to live independently. Barbosa Neves et al¹² demonstrated how digital technologies can enhance social connectedness and support among older adults, reducing loneliness and improving overall well-being.

The reviewed literature strongly supports the positive impact of AI and telehealth on older adults’ health and well-being. These technologies have proven effective in providing accessible, convenient, and personalized healthcare solutions, which are particularly valuable for managing chronic conditions and improving overall health outcomes. However, barriers such as digital literacy and access need to be addressed to ensure equitable benefits across the older adult population. Continued research and targeted interventions are essential to fully realize the potential of these technologies in enhancing the quality of life for older adults.

Barriers for Older Adults

Many older adults may have limited experience with digital technologies,leading to challenges in adopting telehealth and AI tools.

Technological literacy and usability issues

Barnard et al¹³ highlighted the challenges older adults face in learning and using new technologies, such as perceived difficulties and usability issues. Lee et al¹⁴developed the Older Adult Technology Acceptance Model (OATAM) to understand factors influencing older adults’ use of consumer technology, emphasizing the need for user-friendly designs. Vaportzis et al¹⁵ also stressed the importance of early user involvement and iterative testing to enhance technology adoption among older adults.

Privacy and security concerns

Satariano et al¹⁶ discussed the privacy and security concerns associated with health technologies, which can be significant barriers to adoption. Berridge et al¹⁷ examined the ethical implications of surveillance technologies in nursing homes, highlighting the need for robust privacy protections. de Veer et al¹⁸ and Moreno et al¹⁹ further explored the determinants of e-health adoption and the importance of addressing privacy and security issues to gain older adults’ trust.

Cost and economic factors

Peek et al²⁰ and Robinson et al²¹ reviewed the economic factors influencing the adoption of assistive technologies among older adults. Moreira et al.²² discussed the cost implications of telehealth services and clinical decision support systems, stressing the need for affordable solutions. Gell et al²³ investigated technology use patterns among older adults with disabilities, underscoring the economic barriers to technology adoption.

The literature highlights that while telehealth and AI technologies offer significant opportunities for older adults, including improved accessibility, health monitoring, and independence, several barriers must be addressed. These include technological literacy, privacy and security concerns, and economic factors. Addressing these barriers through user-centered design, robust privacy protections, and affordable solutions can enhance the adoption and utilization of these technologies by older adults.

This paper differs from existing studies mainly in the following 3 aspects.

First, in this paper, the adoption of intelligent medical and health care products by doctors in elderly care fields is considered. A 3-party evolutionary game model among the government, intelligent medical care enterprises, and doctors is constructed. The strategic stability of game parties, the effects of various factors on strategy selection, and, in particular, the strategies used by doctors adopting intelligent medical and health care products are analyzed. Secondly, in this paper, stability analysis on the pure and hybrid strategies of replicator dynamics systems is performed using the Lyapunov indirect method, thus obtaining the combination of evolutionary stability strategies under different conditions. Finally, simulation analysis is conducted using MATLAB 2019a to validate the effectiveness of model analysis under different initial conditions. Based on the conclusions of this analysis, countermeasures and suggestions are offered regarding the perfection of the supervision mechanism by the government, the active operations of intelligent medical and health care products by enterprises, and the adoption of intelligent medical and health care products by doctors.

Modeling Assumptions and Construction

Government support is key to guaranteeing the successful launch of intelligent medical and health care products. The active operations of enterprises constitute the premise for ensuring the quality of intelligent medical and health care products. The adoption of intelligent medical and health care products by doctors is the goal to be achieved by the leap in elderly care services. The logical structure of the constructed 3-party evolutionary game model for the adoption of intelligent medical and health care products is shown in Figure 1.

Figure 1.

Logical relationship underlying the 3-party evolutionary game model.

Model Assumptions

The following assumptions underly the construction of the game model, and are used to analyze the strategic and equilibrium point stability of game parties and the effects of various factors on strategy selection:

Assumption 1: In the game model for the adoption of intelligent medical and health care products by doctors, x, y, and z denote government departments, AI medical care enterprises, and doctors, respectively. All 3 parties are output subjects with bounded rationality. Strategy selection employs a gradual tendency towards a stable optimal strategy over time.

Assumption 2: There are 2 strategic spaces for the attitudes of government elderly care departments towards AI medical and health care products: $α = (α_{1}, α_{2}) =$ (support/lack of support of AI medical and health care products). If the government supports AI medical and health care products, it will positively manage and guide intelligent medical care enterprises through rewards and punishments. However, intelligent medical and health care products are emerging products on the market, making it difficult for the government to control their positive effects and negative consequences in advance. That is, government departments act only when intelligent medical and health care products have deviated from expected effects or have caused accidents. This action usually lags behind. When the government vigorously supports intelligent medical and health care products, unqualified or weak AI enterprises will also seek to enter the intelligent medical and health care market, trying to defraud government subsidies with shell products. In the long run, this situation will harm the immediate interests of elderly people and jeopardize the reputation of both the government and the intelligent medical care industry, leading to a loss of government credibility. Conversely, if the government does not support intelligent medical and health care products, the enthusiasm of AI medical care enterprises to invest in R&D and inputs will be weakened, thus causing the whole industry to fall into a slump.

Assumption 3: The attitudes of AI medical care enterprises towards intelligent medical and health care products include: $β = (β_{1}, β_{2})$ , (where $β_{1}$ represents active operations and $β_{2}$ represents passive operations). If intelligent medical care enterprises actively operate intelligent medical and health care products, they will answer the call of the government, and approach R&D pragmatically. In this way, they not only provide high-quality medical and health care products for both elderly people and doctors, they also offer targeted free training and teaching for inexperienced doctors. In addition, they can also predict possible risks in product operations, reserve adequate security deposits, assume responsibility for negative consequences of products, and assist doctors in resolving medical disputes. On the contrary, if intelligent medical care enterprises hold a passive attitude towards the operations of intelligent medical and health care products, they will still act to win the fiscal subsidies the government provides to high-tech enterprises. However, limited by their strength, their passive attitude in dealing with the government and doctors will ultimately cause doctors to resist intelligent medical and health care products.

Assumption 4: Doctors’ attitudes towards intelligent medical and health care products can be characterized as: $γ = (γ_{1}, γ_{2}) =$ (adoption/non-adoption). Doctors will embrace intelligent medical and health care products if government departments vigorously promote them among doctors. In this promotion, the responsibilities for product-related accidents should be explicitly divided, relevant laws and regulations as well as policies should be actively established, and the legitimate rights and interests of doctors must be effectively safeguarded. Doctors will be even more proactive in adopting intelligent medical and health care products if intelligent medical care enterprises provide constant product training, standardize operating steps, take responsibility for operating accidents, and dispel doubts of doctors about product use. In contrast, doctors will reject intelligent medical and health care products if they have doubts or a speculative or risk-taking mentality, or if their rights and interests in using intelligent products as auxiliary diagnosis tools cannot be safeguarded. Doctors’ tendency to reject intelligent medical and health care products will further increase if government departments are passive in supporting the trial use of emerging technologies or if doctors are held responsible for product-related accidents.

Assumption 5: In the stage of the initial game among the government, AI medical care enterprises, and doctors, if the government “supports” the use of intelligent medical and health care products, it will select $α_{1}$ by a probability of $x$ , where $x \in [0, 1]$ . Conversely, if the government “does not support” the use of intelligent medical and health care products, it will select $α_{2}$ by a probability of $1 - x$ , where $x \in [0, 1]$ . If AI medical care enterprises employ the strategy of “active operations,” they will select $β_{1}$ by a probability of $y$ , where $y \in [0, 1]$ . Conversely, if they employ the strategy of “passive operations,” they will select $β_{2}$ by a probability of $1 - y$ , where $z \in [0, 1]$ . If doctors adopt AI medical and health care products, they will select $γ_{1}$ by a probability of $1 - z$ , where $z \in [0, 1]$ . Conversely, if they reject AI medical and health care products, they will select $γ_{2}$ by a probability of $1 - z$ , where $z \in [0, 1]$ .

The payoff matrix considering the 3-party game among the government, enterprises, and doctors is shown in Table 1.

Table 1.

Profit and Loss Indexes of Stakeholders.

AI medical care enterprises	Doctors	Government departments
AI medical care enterprises	Doctors	Support $x$	Does not support $1 - x$
AI medical care enterprises
Active operations $y$	Adoption $z$	$C_{R 1} + C_{F} + G_{C} - G_{P} - C_{1}$	$C_{R 1} - G_{P} + C_{F} - C_{1} - C_{E}$
			$D_{M} - D_{T 1} - D_{V 2} - G_{U}$
		$G_{R} + R_{P} - G_{S 1} + G_{P} - G_{C} - G_{D}$	$- G_{S 2} + G_{P} + G_{U} + G_{R}$
	Non-adoption $1 - z$	$- C_{1} + C_{F} + G_{C} - G_{P}$	$- G_{P} - C_{1} + C_{F}$
		$D_{Q}$	$D_{H}$
		$R_{P} - G_{S 1} + G_{P} - G_{C}$	$- G_{S 2} + G_{P} - D_{H}$
Passive operations $1 - y$	Adoption $z$	$C_{R 2} + G_{C} - G_{P} - C_{2}$	$C_{R 2} - G_{P} - C_{2}$
		$G_{D} - D_{T 2} + D_{V 1}$	$D_{M} - D_{T 2} - D_{V 2} - G_{U}$
		$R_{P} - G_{N} - G_{S 1} - G_{C} - G_{D}$	$- G_{S 2} - G_{N} + G_{P} + G_{U}$
	Non-adoption $1 - z$	$- C_{2} + G_{C} - G_{P}$	$- G_{P} - C_{2}$
		$0$	$0$
		$R_{P} - G_{S 1} - G_{N}$	$- G_{S 2} - G_{N} + G_{P}$

Solving the Evolutionary Game Model

Replicator dynamics equation and equilibrium point analysis

{\begin{array}{l} F (x) = x (1 - x) {G_{s 2} + R_{p} - G_{p} - G_{s 1} - (G_{C} + G_{D} + G_{U}) z + y [D_{H} - G_{C} + G_{p} + (G_{C} - D_{H}) z]} \\ F (y) = y (1 - y) [C_{2} - C_{1} + C_{F} + (C_{R 1} - C_{E} - C_{R 2}) z + C_{E} x z] \\ F (z) = z (1 - z) {D_{M} - D_{T 2} - D_{V_{2}} - G_{V} + (D_{T 2} - D_{H} - D_{T 1}) y + x [y (D_{H} + D_{M} - D_{Q}) + D_{V 1} + D_{V 2} + G_{D} + G_{U} - D_{M}]} \end{array}

Equilibrium points can be obtained from $F (x) = F (y) = F (z) = 0$ , then have

\begin{array}{l} E_{1} (0, 0, 0), E_{2} (1, 0, 0), E_{3} (0, 1, 0), E_{4} (0, 0, 1), E_{5} (1, 1, 0), E_{6} (1, 0, 1), E_{7} (0, 1, 1), E_{8} (1, 1, 1), \\ E_{9} (0, \frac{D_{M} - D_{T 2} - D_{V 2} - G_{U}}{D_{H} + D_{T 1} - D_{T 2}}, \frac{C_{2} + C_{F} - C_{1}}{C_{e} - C_{R 1} + C_{R 2}}), \\ E_{10} (1, \frac{D_{T 2} - D_{V 1} - G_{D}}{D_{M} - D_{Q} - D_{T 1} + D_{T 2}}, \frac{C_{1} - C_{2} - C_{F}}{C_{R 1} - C_{R 2}}), \\ E_{11} (\frac{D_{M} - D_{T 2} - D_{V 2} - G_{U}}{D_{M} - D_{V 1} - D_{V 2} - G_{D} - G_{U}}, 0, \frac{G_{S 2} + R_{P} - G_{P} - G_{S 1}}{G_{C} + G_{D} + G_{U}}), \\ E_{12} (\frac{D_{H} - D_{M} + D_{T 1} + D_{V 2} + G_{U}}{D_{H} - D_{Q} + D_{V 1} + D_{V 2} + G_{D} + G_{U}}, 1, \frac{D_{H} - G_{C} - G_{S 1} + G_{S 2} + R_{P}}{D_{H} + G_{D} + G_{U}}), \\ E_{13} (\frac{C_{1} - C_{2} + C_{0} - C_{F} - C_{R 1} + C_{R 2}}{C_{e}}, \frac{G_{C} + G_{D} + G_{p} + G_{S 1} - G_{S 2} + G_{U} - R_{p}}{G_{p}}, 1) \end{array}

Among these 14 equilibrium points, $E_{1} ~ E_{8}$ are asymptotically stable, while $E_{9} ~ E_{13}$ are non-asymptotically stable; therefore, it is only necessary to discuss the asymptotic stability of $E_{1} ~ E_{8}$ . A Jacobian matrix $J$ is constructed for the replicator dynamics equations of the 3 parties.

J = [\begin{matrix} \partial F (x) / \partial x & \partial F (x) / \partial y & \partial F (x) / \partial z \\ \partial F (y) / \partial x & \partial F (y) / \partial y & \partial F (y) / \partial z \\ \partial F (z) / \partial x & \partial F (z) / \partial y & \partial F (z) / \partial z \end{matrix}]

where

\begin{array}{l} \partial F (x) / \partial x = (1 - 2 x) {G_{S 2} + R_{P} - G_{P} - G_{S 1} - (G_{C} + G_{D} + G_{U}) z + y [D_{H} - G_{C} + G_{P} + (G_{C} - D_{H}) z]}, \\ \partial F (x) / \partial y = x (1 - x) [D_{H} - G_{C} + G_{P} + (G_{C} - D_{H}) z], \\ \partial F (x) / \partial z = x (1 - x) {- (G_{C} + G_{D} + G_{U}) + y (G_{C} - D_{H})}, \\ \partial F (y) / \partial x = y (1 - y) C_{E} z, \\ \partial F (y) / \partial y = (1 - 2 y) [C_{2} - C_{1} + C_{F} + (C_{R 1} - C_{E} - C_{R 2}) z + C_{E} x z], \\ \partial F (y) / \partial z = (1 - 2 y) (C_{R 1} - C_{E} - C_{R 2} + C_{E} x), \\ \partial F (z) / \partial x = z (1 - z) [D_{V 1} - D_{M} + D_{V 2} + G_{D} + G_{U} + (D_{H} + D_{M} - D_{Q}) y], \\ \partial F (z) / \partial y = z (1 - z) [(D_{T 2} - D_{H} - D_{T 1}) + x (D_{H} + D_{M} - D_{Q})], \\ \partial F (z) / \partial z = (1 - 2 z) {\begin{cases} D_{M} - D_{T 2} - D_{V 2} - G_{U} + (D_{T 2} - D_{H} - D_{T 1}) y \\ + x [D_{V 1} - D_{M} + D_{V 2} + G_{D} + G_{U} + (D_{H} + D_{M} - D_{Q}) y] \end{cases}} . \end{array}

The eigenvalues of this Jacobian matrix corresponding to $E_{1} ~ E_{8}$ are solved. According to the Lyapunov theorem, these are stable points of the system when the eigenvalues are all real numbers smaller than 0, and instable points when the eigenvalues are all real numbers greater than 0. Various equilibrium points and their eigenvalues are provided in Table 2.

Table 2.

Analysis of Equilibrium Points.

Equilibrium point	$λ_{1}$	$λ_{2}$	$λ_{3}$
$E_{1} (0, 0, 0)$	$G_{S 2} + R_{p} - G_{P} - G_{S 1}$	$C_{2} + C_{F} - C_{1}$	$D_{M} - D_{T 2} - D_{V 2} - G_{U}$
$E_{2} (1, 0, 0)$	$G_{P} + G_{S 1} - G_{S 2} - R_{p}$	$C_{2} + C_{F} - C_{1}$	$D_{V 1} + G_{D} - D_{T 2}$
$E_{3} (0, 1, 0)$	$D_{H} + G_{S 2} + R_{p} - G_{C} - G_{S 1}$	$C_{1} - C_{2} - C_{F}$	$D_{M} - D_{T 1} - D_{V 2} - G_{U} - D_{H}$
$E_{4} (0, 0, 1)$	$G_{S 2} + R_{p} - G_{C} - G_{D} - G_{P} - G_{S 1} - G_{U}$	$C_{2} + C_{F} + C_{R 1} - C_{1} - C_{E} - C_{R 2}$	$D_{T 2} + D_{V 2} + G_{U} - D_{M}$
$E_{5} (1, 1, 0)$	$G_{C} + G_{S 1} - G_{S 2} - R_{p} - D_{H}$	$C_{1} - C_{2} - C_{F}$	$D_{M} + D_{V 1} + G_{D} - D_{Q} - D_{T 1}$
$E_{6} (1, 0, 1)$	$G_{C} + G_{D} + G_{P} + G_{S 1} + G_{U} - G_{S 2} - R_{p}$	$C_{2} + C_{F} + C_{R 1} - C_{R 2} - C_{1}$	$D_{T 2} - D_{V 1} - G_{D}$
$E_{7} (0, 1, 1)$	$G_{S 2} + R_{p} - G_{C} - G_{D} - G_{S 1} - G_{U}$	$C_{1} + C_{E} + C_{R 2} - C_{2} - C_{F} - C_{R 1}$	$D_{H} + D_{T 1} + D_{V 2} + G_{U} - D_{M}$
$E_{8} (1, 1, 1)$	$G_{C} + G_{D} + G_{S 1} + G_{U} - G_{S 2} - R_{p}$	$C_{1} + C_{R 2} - C_{2} - C_{F} - C_{R 1}$	$D_{Q} + D_{T 1} - D_{V 1} - G_{D} - D_{M}$

Scenario 1: Government departments tend to be non-supportive, AI medical care enterprises tend to engage in passive operations, and doctors tend towards adoption. In this scenario, the equilibrium point $E_{1} (0, 0, 0)$ is the ESS, in which case there must be $G_{S 2} + R_{p} - G_{p} - G_{S 1} < 0$ , $C_{2} + C_{F} - C_{1} < 0$ , $D_{M} - D_{T 2} - D_{V 2} - G_{U} < 0 .$

Scenario 2: Government departments tend to be supportive, AI medical care enterprises tend to engage in passive operations, and doctors tend towards non-adoption. In this scenario, the equilibrium point $E_{2} (1, 0, 0)$ is the ESS, in which case there must be $G_{p} + G_{S 1} - G_{S 2} - R_{p} < 0$ , $C_{2} + C_{F} - C_{1} < 0$ , $D_{V 1} + G_{D} - D_{T 2} < 0$ .

Scenario 3: Government departments tend to be non-supportive, AI medical care enterprises tend to engage in active operations, and doctors tend towards non-adoption. In this scenario, the equilibrium point $E_{3} (0, 1, 0)$ is the ESS, in which case there must be $D_{H} + G_{S 2} + R_{p} - G_{C} - G_{S 1} < 0$ , $C_{1} - C_{2} - C_{F} < 0$ , $D_{H A} - D_{T 1} - D_{V 2} - G_{U} - D_{H} < 0$ .

Scenario 4: Government departments tend to be non-supportive, AI medical care enterprises tend to engage in passive operations, and doctors tend towards adoption. In this scenario, the equilibrium point $E_{4} (0, 0, 1)$ is the ESS, in which case there must be $G_{S 2} + R_{p} - G_{C} - G_{D} - G_{p} - G_{S 1} - G_{U} < 0,$ $C_{2} + C_{F} + C_{R 1} - C_{1} - C_{E} - C_{R 2} < 0,$ $D_{T 2} + D_{V 2} + G_{U} - D_{M}$ $< 0 .$

Scenario 5: Government departments tend to be supportive, AI medical care enterprises tend to engage in active operations, and doctors tend towards non-adoption. In this scenario, the equilibrium point $E_{5} (1, 1, 0)$ is the ESS, in which case there must be $G_{C} + G_{S 1} - G_{S 2} - R_{p} - D_{H} < 0$ , $C_{1} - C_{2} - C_{F} < 0$ , $D_{M} + D_{V 1} + G_{D} - D_{Q} - D_{T 1} < 0$ .

Scenario 6: Government departments tend to be supportive, AI medical care enterprises tend to engage in passive operations, and doctors tend towards adoption. In this scenario, the equilibrium point $E_{6} (1, 0, 1)$ is the ESS, in which case there must be $G_{C} + G_{D} + G_{p} + G_{S 1} + G_{U} - G_{S 2} - R_{p} < 0,$ $C_{2} + C_{F} + C_{R 1} - C_{R 2} - C_{1} < 0$ , $D_{T 2} - D_{V 1} -$ $G_{D} < 0$ .

Scenario 7: Government departments tend to be non-supportive, AI medical care enterprises tend to engage in active operations, and doctors tend towards adoption. In this scenario, the equilibrium point $E_{7} (0, 1, 1)$ is the ESS, in which case there must be $G_{S 2} + R_{p} - G_{C} - G_{D} - G_{S 1} - G_{U} < 0,$ $C_{1} + C_{E} + C_{R 2} - C_{2} - C_{F} - C_{R 1} < 0, D_{Q} + D_{T 1} - D_{V 1} -$ $G_{D} -$ $D_{M} < 0$ .

Scenario 8: Government departments tend to be supportive, AI medical care enterprises tend to engage in active operations, and doctors tend towards adoption. In this scenario, the equilibrium point $E_{8} (1, 1, 1)$ is the ESS, in which case there must be $G_{C} + G_{D} + G_{S 1} + G_{U} - G_{S 2} - R_{p} < 0$ , $C_{1} + C_{R 2} - C_{2} - C_{F} - C_{R 1} < 0$ , $D_{Q} + D_{T 1} - D_{V 1} - G_{D} -$ $D_{M} < 0$ .

Parameter Settings

Initial values are assigned to the parameters in the payoff matrix based on the attitude of the government towards intelligent elderly care, as well as the attitudes of intelligent medical and health care enterprises and doctors towards intelligent medical and health care products, as shown in Table 3.

Table 3.

Parameter Assignments.

$G_{S 1}$	$G_{D}$	$G_{R}$	$R_{P}$	$G_{N}$	$G_{P}$	$G_{C}$	$G_{S 2}$	$G_{U}$	$G_{R 1}$	$C_{1}$
10	10	20	40	25	40	10	8	15	80	80
$C_{F}$	$G_{R 2}$	$C_{2}$	$C_{E}$	$D_{T 1}$	$D_{M}$	$D_{Q}$	$D_{T 2}$	$D_{V 1}$	$D_{V 2}$	$D_{H}$
30	70	70	20	30	40	50	10	40	50	20

The construction of a system of factors influencing the adoption of intelligent medical and health care products by doctors is a project considering multiple influencing subjects. The adoption and popularization of AI medical and health care products cannot be accomplished by doctors alone. It also relies on the co-participation of the government and intelligent medical and health care enterprises. MATLAB 2019a software is used to analyze the evolutionary trajectories of the stability strategies taken by the 3 subjects (ie, the government, AI medical and health care enterprises, and doctors) based on the replicator dynamics equations for the evolutionary game among influencing subjects. The strategies taken by influencing subjects under various scenarios are visualized. Progressive evolutionary trajectories are used to further clarify the evolutionary processes of influencing subjects under different levels of initial willingness. In the schematic diagram of evolutionary trajectories, the $y$ -axis denotes the probabilities (within the interval of [0,1] with which the government supports AI medical and health care products, AI medical and health care enterprises engage in active operations, and doctors adopt AI medical and health care products. The $x$ -axis denotes the evolutionary range $t$ . The preset simulation period is 2, the time step is 2, and the influencing factors between subjects are $x = y = z = 0.5$ .

Simulation Analysis of Evolutionary Game Among 3 Subjects

Simulation analysis of the evolutionary game behavior of the government’s attitude

Based on the simulation assignments of government departments, in this paper, the effects of parameter such as $G_{S 1}$ , $G_{S 2}$ , $G_{C}$ , $G_{P}$ , $G_{D}$ , $G_{U}$ , $R_{P}$ , and $D_{H}$ on the behaviors of enterprises and doctors are analyzed. According to Figure 2a, other things being equal, the threshold of the cost borne by the government for supporting the use of AI medical and health care products is between 10 and 70. If $G_{S 1} \leq 10$ , the government chooses the supportive strategy, suggesting that the behavior strategy of the government is greatly influenced by costs. On the one hand, if the government finds that intelligent medical and health care products are affected by market regulation and that these are highly praised and widely welcomed by doctors and elderly patients once launched, the government can make a small investment and gain a high market return. The high popularity of products also effectively stimulates public consumption, directly contributes to local GDP, and indirectly drives economic growth. By filling the role of a perfect intermediary between doctors and enterprises, government elderly care departments can therefore not only achieve local GDP growth at little cost, it can also gain returns in the forms of career development (because of the linkage between the post promotion of officials and the “economic benefits” achieved during their tenure), economic payback, and government reputation. In this case, the government will be willing to support the use of intelligent medical and health care products. On the other hand, even in the fact of poor market feedback and doctors and elderly patients being unwilling to adopt new technologies, the government can still choose to invest in intelligent medical and health care products thus encouraging the development of new technologies and the growth and expansion of new enterprises. Considering the low trial-and-error costs, the government still has a high willingness to support the use of intelligent medical and health care products. If $G_{S 1} \geq 70$ and the “sunk cost” is sizeable for government departments, they will consider the “input-output ratio.” If the internal rate of return on intelligent medical and health care products is far lower than the rate of return on investment, government departments will choose the non-supportive strategy. If $G_{S 1}$ fluctuates between 10 and 70, the government’s strategy is unstable. The indecisiveness of the government is mainly attributable to a fast cost increase, which produces large disagreements among decision-makers thus making it difficult to reach an internal consensus. The above conclusions are consistent with Figure 2b and c. With increasing $G_{S 2}$ , the time required to evolve to 1 is shortened. In that case, governmental departments will perceive the increase in the “sunk cost” and the failure of market feedback to meet the expectation. Consequently, they will find that the economic growth and government reputation gained cannot compensate for the cost of government input. As shown in Figure 2c, other things being equal, the threshold of $G_{C}$ ranges between 10 and 70. If $G_{C} \leq 10$ , the government selects the supportive strategy and provides subsidies to AI medical care enterprises. If $G_{C} \geq 70$ , the government selects the non-supportive strategy. If $G_{C} = 35$ , there is no stable strategy for the government, which can be explained by the “entropy” theory in informatics: If all samples have equal probability, the entropy value is the largest. If government subsidies to AI medical care enterprises present an equal probability distribution, $G_{C} = 35$ has the highest uncertainty. In summary, the greater the value of $G_{C}$ , the less conducive it is for government departments to select the supportive strategy. It can be concluded from the data shown in Figure 2 that, the higher the cost of input in AI medical and health care products, the less conducive it is for government departments to select the supportive strategy.

Figure 2.

Effects of the government’s attitude on the behavior strategies of game subjects: (a) simulation analysis of the effect of changed $G_{S 1}$ on the strategy selection of governmental departments, (b) simulation analysis of the effect of changed $G_{S 2}$ on the strategy selection of governmental departments, (c) simulation analysis of the effect of changed $G_{C}$ on the strategy selection of governmental departments, (d) simulation analysis of the effect of changed $G_{P}$ on the strategy selection of governmental departments, (e) simulation analysis of the effect of changed $G_{D}$ on the strategy selection of governmental departments, (f) simulation analysis of the effect of changed $G_{U}$ on the strategy selection of governmental departments, (g) simulation analysis of the effect of changed $R_{P}$ on the strategy selection of governmental departments, and (h) simulation analysis of the effect of changed $D_{H}$ on the strategy selection of governmental departments.

As shown in the partially enlarged drawing in Figure 2d, with increasing $G_{P}$ , the time required to evolve to 1 is shortened. According to a comparison with Figure 2f, enterprises are very sensitive to the severity of punishment. If the severity of punishment is 10, the evolutionary path is relatively gentle. If the severity of punishment increases to 90, the slope of the evolutionary path increases abruptly, resulting in a shortened time of evolution to the equilibrium point. This suggests that, in the absence of government supervision, the production cost of AI medical and health care products is a purely economic issue for enterprises. That is, enterprises will strive to reduce R&D costs and shorten the time to market, regardless of product quality. Enterprises will try to defraud government subsidies without any substantial improvement. In this case, punitive measures, especially severe ones, help to curb corporate violations. Threatened by the risk of bankruptcy liquidation, enterprises will carry out compliance R&D as required by the government. As shown in Figure 2f, with the gradual growth of $G_{U}$ , the time required to evolve to 1 is shortened. In contrast to Figure 2d, Figure 2f shows that if the government increases punishment severity (assignment of $G_{U}$ increases from 2 to 18), the evolutionary path increases slowly in succession. This development suggests that doctors are not sensitive to the severity of punishment imposed by the government. This phenomenon can be explained by the professional ethics of doctors. As medical professionals, doctors have no adequate understanding of the principles of intelligent medical and health care products and are consequently distrustful of related decision-making results, and they tend to not adopt such products. Therefore, the severity of punishment has no direct effect on the adoption of intelligent medical and health care products by doctors.

The next step is to study the effect of governmental subsidies on the adoption behavior strategy of doctors. As shown in Figure 2e and h, with increasing $G_{D}$ and the provision of training subsidies by the government to doctors in a targeted manner, the evolution of doctors’ adoption behavior can be accelerated. Consequently, the time required to evolve to the stable equilibrium point in doctor training is shortened. Differing from profit-driven enterprises, doctors deciding to take part in training are largely driven by a thirst for knowledge about information technology and an intention to improve work efficiency and diagnosis quality. This means that, if the government offers suitable motivation or reasonable compensation to doctors who actively take part in training, improve their status, and strengthen their beliefs, doctors’ willingness to adopt intelligent medical and health care products can be promoted. Therefore, with increasing $D_{H}$ , the time required to evolve to 1 is shortened. This suggests that the greater the value of $D_{H}$ , the more conducive it is for government departments to select the supportive strategy.

With the continuous progress of AI technologies, the notion of intelligent medical and health care products and services is being embraced by more and more doctors and patients. As the central government constantly adjusts the indexes and modes by which the performance of local governments is appraised, local government officials have clearer responsibilities to conduct new-type technological innovation. Productive innovation ability is also gradually shifting from the central to the local level, as shown in Figure 2g. As a result, it is difficult to sustain the traditional extensive development model, and local government officials must stress the innovation and development of new technologies to realize their political ambitions. This is certainly a complex process that must be constantly adjusted based on realities under governmental guidance and enterprise cooperation. Therefore, with increasing $R_{P}$ , the time required to evolve to 1 is shortened. This suggests that the greater the value of $R_{P}$ , the more conducive it is for government departments to select the supportive strategy.

Simulation Analysis of the Evolutionary Game Behavior of AI Enterprises’ Attitudes

The next step is to identify how AI enterprises’ attitudes affect the behavioral strategies of participatory subjects. As clearly shown in Figure 3a, with increasing $C_{R 1}$ , the return gained by enterprises, both from active or passive operations, presents a slowly climbing trend in the initial period of evolution. This is probably because the production of intelligent medical products by enterprises also follows the law of the product life cycle. Slow growth of a product in the initial period suggests that the product is still in the input period. In this period, both doctors and patients have a weak cognition of product information and either choose to adopt the product or take a wait-and-see attitude mainly based on previous experience. However, they do have the psychological need to further understand relevant information and features of the product. Because of technical reasons, a product in the input period cannot be launched at large scale. During this period, enterprises face high costs, the growth of sales volume is slow and profit margins are slim. With the continuous operation of the enterprise, the product gradually gains a foothold in the market and enters the growth period through market expansion. During this period, a significant number of doctors and patients have become interested in the product. Certain doctors who used to hold a wait-and-see attitude are gradually persuaded by their peers who have used the product and rated it highly, and initially hesitant doctors also join in. This explains why the latter part of the evolution curve shows acceleration, why the time required to evolve to 1 is shortened, and why it finally converges to 1. In the case of Figure 3a, the more the enterprise invests in active operations, the higher the rate of return the enterprise gains. If the assignment of $C_{R 1}$ is 140, the slope of its evolution curve will rise most rapidly, and the time required to evolve will be the shortest. In the case of Figure 3b, because of the passive operations of the enterprise, doctors lack the trust in intelligent medical and health care products, and the time required to seize market shares is extended. This results in less return for the enterprise compared to that in active operations. Therefore, the greater the value of $C_{R 2}$ , the less conducive it is for AI medical care enterprises to select the active operations strategy.

Figure 3.

Effects of enterprise behaviors on the behavior strategies of game subjects: (a) simulation analysis of the effect of changed $C_{R 1}$ on the strategy selection of enterprises, (b) simulation analysis of the effect of changed on the strategy selection of enterprises, (c) simulation analysis of the effect of changed $C_{1}$ on the strategy selection of enterprises, (d) simulation analysis of the effect of changed $C_{2}$ on the strategy selection of enterprises, (e) simulation analysis of the effect of changed $C_{E}$ on the strategy selection of enterprises, and (f) simulation analysis of the effect of changed $C_{F}$ on the strategy selection of enterprises.

To explore how the change in the attitude of intelligent medical care enterprises towards operations affects their return, the maximum assignments of $C_{1}$ and $C_{2}$ are set to 130 in this paper. As shown in Figure 3c and d, for both the government and enterprises, with increasing cost incurred by the change in attitude towards operations, the evolutionary rate also increases, and the time required to evolve to the stable equilibrium point is shortened. Apparently, enterprises are greatly affected by the change in attitude towards operations. Other things being equal, the threshold of the cost borne by enterprises in active operations ranges between 90 and 110 If $C_{1} \leq 90$ , AI medical care enterprises choose the active operations strategy, but if $C_{1} \geq 110$ , they choose the passive operations strategy. In this paper, the clinical path and clinical decision support module in the medical system of Epic, the second-largest intelligent medical care enterprise in the US, are used for a case study. The process from the collection of data based on medical advice and patient interviews to the making of “decisions” is a very long information chain. Ideally, each node in the chain is completed by Computerized Physician Order Entry, and doctors only need to obtain the final results. However, in reality, intellectualizing each node in the chain can be extremely expensive in terms of required software and hardware inputs. Because of cost considerations, it is impossible to deploy Epic for general intelligent medical products without limit. This means that it is only possible to collect some “important” data from several major links. However, less data leads to more ambiguous decision-making results. The contradiction between data volume and cost is irreconcilable. In summary, the greater the value of $C_{1}$ , the less conducive it is for AI medical care enterprises to select the active operations strategy; the greater the value of $C_{2}$ , the more conducive it is for AI medical care enterprises to select the active operations strategy.

As shown in Figure 3f, other conditions being equal, the greater the return gained in the form of social reputation, the higher the evolutionary rate (ie, convergence rate) of the enterprise. This is mainly because enterprises provide products and services for society, and a high social reputation helps them attract more doctors and patients and in turn, brings more returns. Currently, the medical market faces fierce homogeneous competition, and the main competitors include Overseas Doctor You, IBM Watson, Tencent Miying, and ET Medical Brain. The reputation of a product is, to a very large extent, an extension of the business philosophy of its manufacturer. This explains why many “time-honored brands” still seize considerable market shares. If an enterprise engages in active operations and becomes involved in the decision-making processes of doctors and patients, the enterprise not only reduces the chance of being eliminated, but also creates tax revenue for the government, thus aligning its interests with the interests of the government in certain aspects. Intelligent medical care enterprises of this type tend to be recognized by the government and succeed in promoting their brand image. Therefore, with increasing $C_{F}$ , the time required to evolve to 1 is shortened. This suggests that the greater the value of $C_{F}$ , the more conducive it is for AI medical care enterprises to select the active operations strategy. Similar to the case of $C_{F}$ , with increasing $C_{E}$ , the time required to evolve to 1 is prolonged. This suggests that, the greater the value of $C_{E}$ , the less conducive it is for AI medical care enterprises to select the active operations strategy.

Simulation Analysis of the Evolutionary Game Behavior of Doctors’ Attitudes

Other conditions being equal, if the attitude of AI medical care enterprises shifts from active operations to passive operations, the threshold of $D_{T 1}$ ranges between 40 and 50. If $D_{T 1} \leq 40$ , doctors select the adoption strategy and if $D_{T 1} \geq 50$ , doctors select the non-adoption strategy. This suggests that the greater the value of $D_{T 1}$ , the less conducive it is for doctors to select the adoption strategy. With increasing $D_{T 2}$ , the time required to evolve to 1 is prolonged, suggesting that the greater the value of $D_{T 2}$ , the less conducive it is for doctors to select the adoption strategy. Because intelligent medical products and services mainly target doctors and because patient-oriented products are mostly health monitoring equipment and bedside equipment, doctors naturally become the main users of intelligent medical products. Doctor-oriented auxiliary diagnosis tools and intelligent medical imaging products are largely developed for special departments or special diseases by information technology companies, such as IBM Watson and ET Medical Brain. The developers of these companies are mainly computer professionals. By contrast, doctors know little about the R&D process, have no adequate understanding of relevant computational processes, and are often distrustful of related decision-making results. Therefore, doctors must spend time on product training, independent of whether enterprises engage in active operations or not. Considering that doctors must devote most of their energy to hospitals and patients, spending spare time on development training beyond existing knowledge systems will make it difficult for doctors to make adoption decisions.

Other conditions being equal, when the attitude of the government shifts from support to non-support, the threshold of $D_{V 1}$ ranges between 25 and 35. If $D_{V 1} \leq 25$ , doctors select the non-adoption strategy and if $D_{V 1} \geq 35$ , doctors select the adoption strategy. This suggests that the greater the value of $D_{V 1}$ , the more conducive it is for doctors to select the adoption strategy. Currently, across the world, intelligent medical products (excluding health monitoring systems and wearable devices, which have narrow scopes of application) are mainly doctor-oriented auxiliary diagnostic and treatment products. These products mostly have a long R&D period and depend on the support of human, material, and information resources. This means that for various enterprises, it will be difficult to allocate these resources on such a large scale, and that the government needs to play an active role in promoting and supporting the allocation of resources. If the government fails to provide adequate support and passes on the practical problems related to funds, manpower, and resource allocation to enterprises, enterprises will feel powerless in R&D and in turn offer subpar products to doctors. Ultimately, doctors will find it difficult to place any trust in the quality of data training and will reject such products. Similarly, if the government does not support the use of intelligent medical products, invests little in such products, and leaves all difficulties to enterprises, it will be impossible for enterprises to solve these problems. Although enterprises can employ an active operations strategy, they still must bear speculative costs, such as false advertisement and public relations spending when dealing with doctors, to persuade doctors to use their products. The cost borne by doctors in the form of mental pressure for accepting corporate misconduct is also a factor that influences the adoption of products by doctors. As a result, with increasing $D_{V 2}$ , the time required to evolve to 1 is prolonged, suggesting that the greater the value of $D_{V 2}$ , the less conducive it is for doctors to select the adoption strategy.

When AI medical care enterprises engage in active operations, the change in the attitude of doctors toward products can be seen from a comparison between Figure 4e and f. Other conditions being equal, the threshold of $D_{M}$ ranges between 20 and 30. If $D_{M} \leq 20$ , doctors select the non-adoption strategy and if $D_{M} \geq 30$ , doctors select the adoption strategy. According to Maslow’s hierarchy of needs, the government can satisfy doctors’ physiological needs by guaranteeing a considerable legitimate income to them. The government can also meet their safety needs by cracking down on gang crimes, strengthening the efforts to combat “medical disturbances,” and aggravating the punishment against “injuries to medical staff.” AI enterprises can satisfy doctors’ social needs by engaging in active operations, narrowing the income gap between various doctors, and establishing a harmonious relationship between hospitals as well as between doctors. AI medical products can meet doctors’ respect needs and self-realization needs by allowing them to dispense with excessive diagnosis and treatment, improving patient satisfaction, and giving a spiritual return to doctors. Therefore, the greater the value of $D_{M}$ , the more conducive it is for doctors to select the adoption strategy. Contrary to $D_{M}$ , $D_{Q}$ has a threshold ranging between 60 and 70. If $D_{Q} \leq 60$ , doctors select the adoption strategy if $D_{Q} \geq 70$ , doctors select the non-adoption strategy. In the case where the government supports the use of intelligent medical products and enterprises engage in active operations, doctors refusing to use intelligent medical products will not only bear the dual pressure from the government and enterprises, they will also face a series of problems incurred by the failure to use intelligent medical products. These roblems include low work efficiency, high misdiagnosis rate, excess patient waiting time, and reduced service quality. These factors will combine and ultimately compromise the reputation of doctors, subjecting them under strong mental pressure. Therefore, the greater the value of $D_{Q}$ , the less conducive it is for doctors to select the adoption strategy.

Figure 4.

Effects of doctors’ behavior on the behavioral strategies of game subjects: (a) simulation analysis of the effect of changed $C_{T 1}$ on the strategy selection of doctors, (b) simulation analysis of the effect of changed $C_{T 2}$ on the strategy selection of doctors, (c) simulation analysis of the effect of changed $C_{V 1}$ on the strategy selection of doctors, (d) simulation analysis of the effect of changed $C_{V 2}$ on the strategy selection of doctors, (e) simulation analysis of the effect of changed . $D_{M}$ on the strategy selection of doctors, (f) simulation analysis of the effect of changed $D_{Q}$ on the strategy selection of doctors, (g) simulation analysis of the effect of changed $D_{H}$ on the strategy selection of doctors, and (h) simulation analysis of the effect of changed $G_{U}$ on the strategy selection of doctors.

When the government does not support the use of AI medical and health care products and doctors refuse to adopt them, the compensation offered by the government to doctors is shown in Figure 4g. With increasing $D_{H}$ , the time required to evolve to 1 is shortened, suggesting that the greater the value of $D_{H}$ , the more conducive it is for doctors to select the adoption strategy. To prevent corruption, doctors shift from adoption to rejection of AI medical and health care products, which may reduce doctors’ turnover rate, lower the quality of diagnosis and treatment, and greatly impact doctors’ performance appraisal. Therefore, the greater the value of $D_{H}$ , the more conducive it is for doctors to select the adoption strategy. In contrast, if doctors adopt intelligent medical and health care products in violation of rules, government departments will punish them to maintain social stability and safeguard the safety of elderly patients. Therefore, with increasing $G_{U}$ , the time required to evolve to 1 is prolonged, suggesting that the greater the value of $G_{U}$ , the less conducive it is for doctors to select the adoption strategy.

Conclusions

Scientific Contributions and Novelty

This study presents a pioneering approach to understanding the behavioral evolution of multiple stakeholders in the adoption of AI-based elderly care products through the lens of evolutionary game theory. By constructing a 3-party evolutionary game model involving government departments, AI medical care enterprises, and doctors, the research offers a novel framework for analyzing the stability of strategic interactions among these key players. This approach provides a deeper understanding of the dynamics influencing the adoption of AI medical and healthcare products, which is critical in addressing the aging population crisis. The use of MATLAB 2019a for simulation analysis adds a quantitative dimension that enhances the robustness of the findings.

Implications to Theory and Practice

The implications of this study extend both to theoretical advancements and practical applications:The study contributes to the body of knowledge on evolutionary game theory by applying it in the context of AI healthcare product adoption. It bridges the gap in existing literature by focusing on doctors’ adoption behavior, a factor often overlooked in previous studies that primarily concentrated on patient and hospital perspectives. The identification of 8 evolutionary stability strategies (ESS) offers a comprehensive understanding of how different parameters affect the strategic decisions of stakeholders.²⁴

The findings of this study have significant implications for policymakers, AI medical enterprises, and healthcare practitioners. For policymakers, the study highlights the importance of financial incentives and regulatory frameworks in encouraging the adoption of AI healthcare products. AI enterprises can benefit from understanding the critical role that doctors’ trust and adoption behavior play in the successful deployment of these products. Healthcare practitioners, particularly doctors, can leverage this understanding to better navigate the integration of AI technologies into their practice, ultimately improving patient outcomes.

Key Lessons Learned

Governmental support, particularly in the form of subsidies and policy guidance, is crucial for the widespread adoption of AI medical products. This support not only influences the behavior of AI enterprises but also significantly impacts doctors’ willingness to adopt these technologies.

The attitude of AI enterprises towards active or passive operations greatly affects the adoption of their products. Active operations, characterized by ongoing R&D, training, and support, lead to higher adoption rates among doctors. In contrast, passive operations can result in substandard products and lower adoption rates.

The costs and benefits perceived by doctors play a pivotal role in their decision to adopt AI medical products. Reducing perceived risks and providing adequate training are essential for increasing adoption rates.

Limitations of the Research

While this study provides valuable insights, it has certain limitations that should be acknowledged:The evolutionary game model used in this study is based on several assumptions, such as bounded rationality and fixed payoff matrices, which may not fully capture the complexity of real-world scenarios. Future research could explore more dynamic models that account for changing preferences and external shocks.The study focuses on the adoption behavior of doctors and does not extensively explore the perspectives of patients and other healthcare providers. Future studies could broaden the scope to include these stakeholders, providing a more holistic view of the adoption process.The use of MATLAB 2019a for simulation, while powerful, has certain limitations in terms of scalability and real-time analysis. Future research could employ more advanced simulation tools or real-world data to validate the findings.

This study offers a significant contribution to the understanding of the adoption of AI medical and healthcare products, particularly in the context of elderly care. The findings underscore the importance of governmental support, the role of AI enterprises in ensuring product quality, and the critical influence of doctors’ adoption behavior. By addressing these factors, stakeholders can effectively promote the integration of AI technologies in healthcare, thereby improving the quality of care for the aging population. Future research should continue to build on these insights, exploring new models and broader perspectives to enhance the adoption and implementation of AI in healthcare.

Footnotes

Acknowledgements

We would like to thank Mogo Edit for their assistance with polish language in this research.

Author Contributions

Conceptualization, J.Y.; methodology, J.Y. and X.W.; writing—original draft preparation, J.Y. and X.W.; writing—review and editing, J.Y. and X.W. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The data that support the findings of this study are available from simulation upon reasonable institutional request.

Declaration of Conflicting Interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This paper was supported by Scientific Research Foundation for High-level Talents of Anhui University of Science and Technology (ID: 2024yjrc18).

Ethical Statement

The data in this study comes from simulation data and does not involve personal or animal experiments. This study does not involve ethical issues.

Informed Consent/Patient Consent

Not applicable

ORCID iD

Jinxin Yang

References

Yin

Bai

Zhang

Prevention schemes for future fresh agricultural products (FAPs) supply chain: mathematical model and experience of guaranteeing the supply of FAPs during the COVID-19 pandemic. J Sci Food Agr. 2021;101:6368-6383.

Yin

Dong

Wang

Measuring the impact of the COVID-19 epidemic on university resumption and suggestions for countermeasures. Front Public Health. 2022;10:1037818.

Yin

Zhang

Dong

Preventing COVID-19 from the perspective of industrial information integration: evaluation and continuous improvement of information networks for sustainable epidemic prevention. J Ind Inf Int. 2020;19:100157.

Cheung

van Velden

Lagerburg

Minderman

The organizational and clinical impact of integrating bedside equipment to an information system: a systematic literature review of patient data management systems (PDMS). Int J Med Inform. 2015;84:155-165.

Peek

Wouters

van Hoof

Luijkx

Boeije

Vrijhoef

HJ.

Factors influencing acceptance of technology for aging in place: a systematic review. Int J Med Inform. 2014;83:235-248.

Greenhalgh

Wherton

Shaw

Morrison

Video Consultations for Covid-19. British Medical Journal Publishing Group; 2020.

Dicianno

Parmanto

Fairman

, et al. Perspectives on the evolution of mobile (mHealth) technologies and application to rehabilitation. Phys Ther. 2015;95:397-405.

Husnain

Hussain

Shahroz

Ali

Hayat

A precision health initiative for chronic conditions: design and cohort study utilizing wearable technology, machine learning, and deep learning. Int J Adv Eng Technol Innov. 2024;1:118-139.

Bonoto

de Araújo

Godói

, et al. Efficacy of mobile apps to support the care of patients with diabetes mellitus: a systematic review and meta-analysis of randomized controlled trials. JMIR Mhealth Uhealth. 2017;5:e4.

10.

Kim

Lee

Smart devices for older adults managing chronic disease: a scoping review. JMIR Mhealth Uhealth. 2017;5:e7141.

11.

Sintonen

Immonen

Telecare services for aging people: assessment of critical factors influencing the adoption intention. Comput Hum Behav. 2013;29:1307-1317.

12.

Barbosa Neves

Franz

Judges

Beermann

Baecker

. Can digital technology enhance social connectedness among older adults? A feasibility study. J Appl Gerontol. 2019;38:49-72.

13.

Barnard

Bradley

Hodgson

Lloyd

AD.

Learning to use new technologies by older adults: perceived difficulties, experimentation behaviour and usability. Comput Hum Behav. 2013;29:1715-1724.

14.

Lee

Kim

Choi

. The adoption of virtual reality devices: The technology acceptance model integrating enjoyment, social interaction, and strength of the social ties. Telemat Inform. 2019;39:37-48.

15.

Vaportzis

Giatsi Clausen

Gow

AJ.

Older adults perceptions of technology and barriers to interacting with tablet computers: a focus group study. Front Psychol. 2017;8:1687.

16.

Satariano

Scharlach

Lindeman

Aging, place, and technology: toward improving access and wellness in older populations. J Aging Health. 2014;26:1373-1389.

17.

Berridge

Halpern

Levy

Cameras on beds: the ethics of surveillance in nursing home rooms. AJOB Empir Bioeth. 2019;10:55-62.

18.

de Veer

Peeters

Brabers

Schellevis

Rademakers

Francke

AL.

Determinants of the intention to use e-Health by community dwelling older people. BMC Health Serv Res. 2015;15:1-9.

19.

Moreno

Boldrini

Moreda

Romá-Ferri

MT.

Drugsemantics: a corpus for named entity recognition in spanish summaries of product characteristics. J Biomed Inform. 2017;72:8-22.

20.

Peek

Luijkx

Rijnaard

, et al. Older adults’ reasons for using technology while aging in place. Gerontology. 2016;62:226-237.

21.

Robinson

Gibson

Kingston

, et al. Assistive technologies in caring for the oldest old: a review of current practice and future directions. Aging Health. 2013;9:365-375.

22.

Moreira

Rodrigues

Korotaev

, et al. A comprehensive review on smart decision support systems for health care. Ieee Syst J. 2019;13:3536-3545.

23.

Gell

Rosenberg

Demiris

LaCroix

Patel

KV.

Patterns of technology use among older adults with and without disabilities. Gerontologist. 2015;55:412-421.

24.

Yin

Zhang

Prevention schemes for future pandemic cases: mathematical model and experience of interurban multi-agent COVID-19 epidemic prevention. Nonlinear Dyn. 2021;104:2865-2900.

Navigating the Adoption Maze: Evolutionary Dynamics of Stakeholder Behavior in AI-Driven Elderly Care Solutions

Abstract

Keywords

Introduction

Related Literature

Opportunities and Benefits for Older Adults

Increased accessibility and convenience

Enhanced health monitoring and management

Improved independence and quality of life

Barriers for Older Adults

Technological literacy and usability issues

Privacy and security concerns

Cost and economic factors

Modeling Assumptions and Construction

Model Assumptions

Solving the Evolutionary Game Model

Parameter Settings

Simulation Analysis of Evolutionary Game Among 3 Subjects

Simulation analysis of the evolutionary game behavior of the government’s attitude

Simulation Analysis of the Evolutionary Game Behavior of AI Enterprises’ Attitudes

Simulation Analysis of the Evolutionary Game Behavior of Doctors’ Attitudes

Conclusions

Scientific Contributions and Novelty

Implications to Theory and Practice

Key Lessons Learned

Limitations of the Research

Footnotes

Acknowledgements

Author Contributions

Data Availability Statement

Declaration of Conflicting Interests

Funding

Ethical Statement

Informed Consent/Patient Consent

ORCID iD

References