Use of artificial intelligence and health-related life satisfaction among older adults: A structural equation modeling study

Abstract

Background

As artificial intelligence (AI) tools become increasingly integrated into healthcare, AI tools support disease management and well-being in older population. However, adoption of AI technologies is often hindered in the population, raising questions about how technology acceptance translates into direct health benefits for older adults.

Objectives

Guided by the Technology Acceptance Model, this study aims to examine the structural mechanisms by which AI healthcare technology variables, including AI competency, attitude, or use frequency, influence health-related life satisfaction among middle- to older-aged adults.

Methods

This study was a secondary analysis using data from 2024 Digital Divide Survey in South Korea. The analytic sample included 582 participants aged ≥ 55 years. Structural equation modeling was used to examine associations among AI competency, attitude, or perceived helpfulness, AI use frequency, and health-related life satisfaction.

Results

The measurement and structural equation models demonstrated acceptable fit. AI competency was positively associated with AI healthcare helpfulness (β = .23, p < .001), AI attitude (β = .53, p < .001), and health-related life satisfaction (β = .28, p < .001). AI attitude was significantly associated with AI healthcare use frequency (β = .29, p < .001), whereas AI healthcare use frequency and helpfulness showed no significant direct effects on health-related life satisfaction.

Conclusions

AI competency, a psychological aspect of AI use, is a more crucial determinant of physical and mental well-being in later life than the increased frequency of AI healthcare use. Interventions should prioritize strengthening AI literacy and self-efficacy among underserved older adults.

Keywords

artificial intelligence technology acceptance model older adults life satisfaction digital divide

Introduction

According to the United Nations, by 2050, one in six people worldwide will be aged 65 years or older, nearly doubling the proportion observed in 2021.¹ Population aging has emerged as one of the most profound demographic trends, with far-reaching implications for health systems, social structures, and overall well-being.² As societies grow older,³ many individuals experience multiple chronic conditions, functional decline, and social isolation, which can reduce life satisfaction and quality of life.

Life satisfaction, a subjective evaluation of one’s overall quality of life across physical, psychological, and social domains, is a critical indicator in aging and health research.³ Higher life satisfaction has been linked to lower morbidity, greater functional independence, and stronger psychological resilience in later life.^4,5 However, many aging individuals report relatively low levels of life satisfaction, which is often associated with declining physical health, reduced social participation, and psychological distress.^6–9 Such low life satisfaction is a significant risk factor for depression, frailty, and reduced longevity.^10–12 Therefore, proactive strategies are needed to enhance well-being among the aging population.^4,13

In this context, artificial intelligence (AI)-driven healthcare technologies are emerging as transformative tools to enhance self-management, disease prevention, and overall quality of life among aging populations.^13,14 AI applications, including predictive analytics, wearable devices, and conversational agents, can support daily health monitoring, medication adherence, and early detection of health risks.^15–17 By providing continuous, data-driven, and personalized feedback, AI systems empower the aging population to actively participate in their own care, potentially improving life satisfaction and well-being.¹⁸ AI technologies relevant to older adults include devices equipped with algorithms for health monitoring, dietary coaching applications that use image recognition, and voice-activated assistants that provide medication reminders or advice.^16,17,19 Frequently integrated into smartphones, smartwatches, and digital health platforms, these tools offer continuous, personalized feedback, supporting self-management and early risk detection. Such technologies represent the AI systems older adults may encounter daily, even without direct clinical intervention.

However, the adoption of AI healthcare technologies among older adults remains limited. Many face barriers such as low digital literacy, usability challenges, data privacy concerns, and skepticism toward AI systems.^4,18 This phenomenon, often referred to as the “gray digital divide,” highlights disparities in digital participation, which can exacerbate health inequalities and social exclusion among older populations.²⁰ Thus, understanding the behavioral and cognitive mechanisms that influence AI acceptance is critical for fostering equitable and sustainable digital health innovation.²¹

In healthcare contexts, the technology acceptance model (TAM) and related frameworks have been widely applied to telehealth, mobile health, and wearable systems to explain technology adoption and engagement among older adults.²² Although previous studies have used TAM to examine technology adoption in this population, most have focused on general technology use or single-device contexts, such as smartphones or telehealth platforms, rather than AI-driven healthcare systems.^23–26 Previous reviews have identified that technology use can enhance quality of life or mental well-being in older adults.^27,28 However, little is known about the mechanisms by which AI healthcare technology acceptance translates into improved life satisfaction and quality of life.

Building on this gap, this study aimed to explore the structural mechanisms underlying AI healthcare technology acceptance, usage behaviors, and their effects on health-related life satisfaction among older adults. The findings are expected to inform the design of AI literacy programs, user-centered healthcare technologies, and policy strategies that promote equitable adoption of digital innovations and enhance life satisfaction and well-being among the rapidly growing aging population.

Methods

Study design and participants

This study used data from the 2024 Digital Divide Survey conducted by the National Information Society Agency of Korea, which is publicly available on the government website (data.go.kr) and did not require additional permission for secondary analysis.²⁹ The purpose of this annual survey is to examine the outcomes of national policies in resolving inequality in digital accessibility in the population of all ages and to provide fundamental information for future policy development. The data collection period was from October to December 2024. The primary data were obtained from a nationally representative sample aged ≥7 years (N = 7,000). From the entire sample, participants aged ≥55 years (n = 2,300) were initially considered, as they were defined as the aging population by the primary data source. Of these, participants who reported awareness of AI healthcare services were considered eligible for the present analysis. Complete-case analysis was then applied, and participants with missing values in the main study variables, including AI healthcare use frequency, were excluded (n = 1,718). Consequently, 582 observations were included in the analysis. The entire data selection process is illustrated in Figure 1.

Figure 1.

Sample flow chart.

Theoretical framework

This study was guided by TAM, which has been widely used to examine human behavior related to technology acceptance. According to the model,³⁰ perceived ease of use is associated with perceived usefulness, both of which can affect attitudes toward using technology and ultimately shape behavioral intentions, leading to actual technology use. The Technology Acceptance Model posits that, among older adults, perceived usefulness and attitude toward AI technologies are key mediators of the relationship between AI competency and the likelihood of adoption.³⁰ Among older adults, affective and cognitive variables, such as AI competency/literacy, self-efficacy, and trust, play particularly influential roles in shaping acceptance.²¹ Based on the literature and theoretical framework, corresponding items were selected from the survey data. The study also extended the original TAM to examine whether AI technology use can affect a health-related outcome—life satisfaction—as an endpoint in the aging population.

AI technologies relevant to older adults include wearable devices equipped with AI algorithms for fall detection and vital sign monitoring, AI-powered dietary coaching applications utilizing image recognition, and voice-activated assistants that offer medication reminders or health advice. Consistent with previous research demonstrating that digital competency and technology-related self-efficacy influence perceived usefulness, behavioral intention, and life satisfaction in later life,^31–33 AI competency was modeled to influence perceived usefulness and attitude, ultimately linking AI use to life satisfaction in older adults. To identify potential intervention targets for improving health-related outcomes through AI technology, this study examined the direct pathways between perceived ease of use or usefulness of AI technology and the health-related outcome (see Figure 2).

Figure 2.

Conceptual framework of the study design

Measurements

The variables of interest were selected based on TAM structures. In addition to theory-based variables, demographic variables regarding participants’ age, sex, and educational level were assessed. Information regarding participants’ AI-related technology accessibility, specifically to various devices such as desktop computers, laptops, smartphones, and past experiences of any AI healthcare use were also examined. Participants’ experiences of previous AI healthcare use were also assessed using a 5-point Likert scale ranging from “never used” to “often.” The following variables were selected to examine the hypothesized relationships in this study.

AI competency. Both perceived ease of use and user competency are highly correlated and are known to strongly predict technology use in the healthcare field.^25,34 Therefore, in this study, AI competency was adopted as a theoretically aligned proxy for perceived ease of use, an extended construct encompassing the essence of ease of use. Twelve items for the latent variable of AI competency were considered. These items address levels of practical proficiency, skillfulness, technical understanding, and ethical responsibility for AI use. Examples of the items include: “I can effectively use AI applications or products in my daily work,” “It is not difficult for me to learn to use new AI applications or products,” and “I understand what kinds of assistance I can receives from AI technologies.” Each item is rated on a 4-point Likert scale ranging from “strongly disagree” to “strongly agree.” A higher score indicates better AI competency for each item.

AI healthcare helpfulness. Perceived usefulness was interpreted as the perceived helpfulness of AI healthcare technology in this study because the item in the survey reflects the perceived benefits of using AI healthcare. A single item was used regarding the level of helpfulness in the respondent’s life of using any AI healthcare technology for exercise coaching, diet solutions, nutrient analysis, including AI photo recognition, prescription analysis, or other mobile applications. In this study, AI healthcare refers to technologies made for consumers that use AI to help people manage their health. For example, there are apps that analyze food from photos, track activity with smart sensors, send health risk alerts, and remind users to take their medication.^16,17,19 These examples come from the Digital Divide Survey’s definition and match tools that are commonly used by older adults in South Korea. In this study, AI healthcare technologies are defined as consumer-facing, low-risk, and non-clinical applications aimed at supporting daily health self-management rather than facilitating clinical decision-making. These include AI-enabled mobile applications and wearable-based services that deliver personalized feedback related to physical activity, dietary habits, and routine health monitoring. The item uses a 4-point Likert scale ranging from “strongly disagree” to “strongly agree.” Higher scores indicate greater helpfulness of AI healthcare technologies.

AI attitude. A total of 10 items were considered as a whole for the underlying construct of attitude toward AI in the sample. The items generally assess overall positive (6 items) and negative (4 items) attitudes toward AI technology. Items assessing negative attitudes were reverse-coded for the analysis. Examples of the items includes: “AI technology will enable the provision of more convenient and better digital services” and “The convenience of AI technology may increase the risk of discrimination or unfair treatment toward specific groups.” Each item is rated on a 4-point Likert scale, with higher scores indicating a positive attitude toward AI technology. Responses included “strongly disagree,” “disagree,” “agree,” and “strongly agree.”

AI healthcare use frequency. A recent systematic review identified inconsistent relationships between behavioral intention and actual use of technology often reported in research, and some of the included studies treated actual use as a more direct outcome.³⁴ Because no direct item considering the user’s behavioral intention was available in the survey, a single item asking about the frequency of use of AI healthcare technologies of any kind during the past month was considered as a proxy indicator of usage intention. We measured how often people use AI in healthcare with a single question. This method is often used in large population surveys that look at general exposure or engagement instead of specific behaviors. Previous studies using the Technology Acceptance Model and digital health research have shown that simple measures of actual use can be practical and valid substitutes for real-world technology engagement in population studies.^23,34,35 The item used in this study rated on a 4-point Likert scale including “never,” “rarely,” “sometimes,” and “often,” with higher scores indicating more frequent use of AI technologies.

Health-related life satisfaction. The original items from the 2024 Korea Digital Divide Survey assessed satisfaction across multiple domains of daily life, including recreational and cultural activities, financial status, social activities, interpersonal relationships, family relationships, work, physical and mental health, and political activities.²⁹ In the present study, the item specifically related to satisfaction with physical and mental health was selected to represent health-related life satisfaction. This item was measured using a 4-point Likert scale. Responses included “never satisfied at all,” “slightly dissatisfied,” “somewhat satisfied,” and “very satisfied.” Higher scores indicate better life satisfaction with physical and mental health.

Data analysis

Stata/SE 17.0 (StataCorp LLC, College Station, TX, USA) was used for descriptive and structural equation modeling analyses. Demographic variables were summarized using descriptive statistics. The entire sample was also divided into two groups: one group included participants who had never used AI healthcare technologies before, and the other included those who had used the technologies at least once. Participants’ information regarding demographic characteristics and AI-related technology accessibility was analyzed using descriptive statistics, including frequencies and percentages. These variables were further compared using chi-square tests of independence according to prior AI healthcare use experience. For descriptive group comparisons, the variable of previous AI healthcare use was converted into a binary variable: “never used” and “used” (combining “rarely,” “sometimes,” and “often”). However, the original frequency scale of AI healthcare use was retained in the structural equation modeling analysis.

To examine the validity and reliability of the measurement model, a two-step approach to structural equation modeling was employed.³⁶ Confirmatory factor analysis was first conducted to validate the latent constructs of AI competency and AI attitude in the measurement model. Several parameters were used to evaluate measurement model fit, such as chi-square, root mean square error of approximation (RMSEA), comparative fit index (CFI), Tucker-Lewis index (TLI), standardized root mean squared residual (SRMR), and coefficient of determination (CD).³⁷ The factor loadings of each item on the underlying constructs and their uniqueness were assessed. Average variance extracted (AVE), and composite reliability (CR), and Cronbach’s α were used to assess convergent validity and reliability. Convergent validity was primarily evaluated using AVE and CR values. AVE values above 0.50 and CR values above 0.70 were considered indicative of acceptable convergent validity.³⁸ However, constructs with AVE values slightly below 0.50 were considered acceptable when CR values were adequate, based on prior methodological recommendations.³⁸ Cronbach’s α values greater than .7 were considered to be acceptable.³⁷ The square root of AVE of each underlying construct was also examined to assess discriminant validity in the model, with a threshold level of 0.7.³⁸ Finally, structural equation modeling was employed to examine the proposed theoretical structures using the same indices for the measurement model. Age, sex, and education level were included as covariates based on prior literature³⁹ and to maintain model parsimony within the structural equation modeling framework. Standardized coefficients (β) were used to examine the path relationships in the model. Significance was set at P< .05. Both measurement and structural equation modeling were performed using the maximum likelihood estimation method.

Ethical considerations

The institutional review board of Ewha Womans University reviewed and approved the study protocol (IRB no. ewha-202509-0014-01). Because this is a secondary analysis of publicly available data, consent was not required, and only de-identified information was obtained.

Results

Sample characteristics

In total, 582 participants were included in the analysis. The mean age of the participants was 63.99 ± 6.72 years, with ages ranging from 55 to 85 years. Slightly more than 50% of the sample were female, and most of the participants had completed high school education, followed by a bachelor’s degree. Almost all participants had no disabilities (95%), and only 12% lived alone. Most participants had access to either a desktop computer (56%) or a smartphone (98%). All participants had internet access. Participants aged <65 years were more likely to report prior AI healthcare use than those aged ≥65 years (χ² = 4.38, p = .036). Higher education level was also significantly associated with prior AI healthcare use (χ² = 29.45, p < .001). Regarding AI-related technology accessibility, participants with access to laptops (χ² = 18.28, p < .001), tablet PCs (χ² = 41.66, p < .001), and smart accessories (χ² = 31.24, p < .001) were more likely to report prior AI healthcare use (see Table 1).

Table 1.

Characteristics of study participants and group differences by AI healthcare use (N=582).

	Total	Never used (n=363)	Used (n=219)	$x^{2}$ (p)
Demographics
Age
<65 years	340 (58.42)	200 (58.82)	140 (41.18)	4.38 (.036)
≥65 years	242 (41.58)	163 (67.36)	79 (32.64)
Sex
Male	288 (49.48)	191 (66.32)	97 (33.68)	3.79 (.052)
Female	294 (50.52)	172 (58.50)	122 (37.63)
Education
Elementary	35 (6.01)	27 (77.14)	8 (22.86)	29.45 (<.001)
Middle school	66 (11.34)	57 (86.36)	9 (13.64)
High school	356 (61.17)	218 (61.24)	138 (38.76)
Bachelor’s	125 (21.48)	64 (48.80)	64 (51.20)
Disability
No	553 (95.02)	341 (64.66)	212 (38.34)	2.37 (.124)
Yes	29 (4.98)	22 (75.86)	7 (24.14)
Living alone
Yes	70 (12.03)	47 (67.14)	23 (32.86)	.77 (.380)
No^a	512 (87.97)	316 (61.72)	196 (38.28)
AI-related Technology Accessibility
Desktop computer
Yes	323 (55.50)	195 (60.37)	128 (39.63)	1.24 (.266)
No	259 (44.50)	168 (64.86)	91 (35.14)
Laptop
Yes	225 (38.66)	116 (51.56)	109 (48.44)
No	357 (61.34)	247 (69.19)	110 (30.81)	18.28 (<.001)
Smartphone
Yes	571 (98.11)	354 (62.00)	217 (38.00)	1.81(.179)
No	11 (1.89)	9 (81.82)	2 (18.18)
Tablet PC
Yes	106 (18.21)	37 (34.91)	69 (65.09)	41.66 (<.001)
No	476 (81.79)	326 (68.49)	150 (31.51)
Smart accessories^b
Yes	89 (15.29)	32 (35.96)	57 (64.04)	31.24 (<.001)
No	493 (84.71)	331 (67.14)	162 (32.86)
Internet
Yes	582 (100)	363 (62.37)	219 (37.63)

Data represented as n (%).

^aLiving with more than two people including family members, relatives, or others.

^bRefers to smart watch, smart health band, AI speakers, and etc.

Measurement model

The goodness-of-fit of the measurement model was acceptable (see Table 2). All observed item information loaded significantly onto their respective latent constructs. The values of the factor loadings and uniqueness of each item for either AI competency or AI attitude were all acceptable. Convergent validity for AI competency was also acceptable, with an AVE value greater than 0.5. Although the AVE value for the attitude construct was slightly below the recommended threshold (AVE = .448), the construct demonstrated acceptable composite reliability (CR=.898), supporting acceptable convergent validity. However, composite reliability and Cronbach’s α were both high (.898 and .888, respectively), which supports good internal consistency and convergent validity.³⁸ Discriminant validity of AI competency was acceptable (square root of AVE = 0.768). Although the discriminant validity estimate for AI attitude was slightly below the recommended threshold (square root of AVE = 0.669), the construct was retained because of its strong composite reliability and theoretical relevance.

Table 2.

Confirmatory factor analysis for measurement model.^a

	Item	FL	Uniqueness	AVE	CR	Cronbach $α$
AI competency	1	.779	.351	.589	.952	.945
	2	.785	.359
	3	.781	.352
	4	.785	.362
	5	.780	.373
	6	.775	.378
	7	.805	.315
	8	.799	.336
	9	.783	.358
	10	.764	.361
	11	.710	.343
	12	.653	.410
AI attitude	1	.671	.528	.448	.898	.888
	2	.655	.493
	3	.673	.526
	4	.629	.531
	5	.686	.511
	6	.639	.540
	7	.600	.589
	8	.656	.515
	9	.743	.420
	10	.736	.413

FL, factor loadings; AVE, average variance extracted; CR composite reliability.

^a $x^{2}$ (208)=682.31 p<.001, root mean square error of approximation =0.063, comparative fit index = 0.936, Tucker-Lewis index = 0.929, standardized root mean squared residual = 0.044, coefficient of determination = 0.992.

Structural equation model

The fit of the hypothesized theoretical model was acceptable, as shown in Figure 3. The structural model exhibited good fit (RMSEA = .055, CFI = .926), meeting conventional thresholds for structural equation models.³⁷ All standardized loadings exceeded 0.60, indicating acceptable item performance. As shown, perceived AI competency had strong positive direct effects on perceived AI helpfulness (β = .23, P <.001) and AI attitude (β = .53, P <.001). AI attitude was also affected by perceived AI helpfulness (β = .21, P <.001), which in turn positively affected frequency of AI healthcare use (β = .29, P <.001). Frequency of AI healthcare use and perceived AI helpfulness did not show any significant direct effects on health-related satisfaction, but perceived AI competency had a positive direct effect on health-related life satisfaction (β = .28, P <.001) in the sample.

Figure 3.

Results of the structural equation modeling analysis^a

Discussion

Given the rapid innovation rate of healthcare technology, an increasing number of older adults are encountering AI-based devices and related technologies in their everyday lives. To better support the aging population and maximize the benefits of AI technologies, gaining a deeper understanding of the mechanisms underlying technology use and its impact on health outcomes is essential. This study examined the theoretical structures of AI technology use among middle-aged and older adults and further explored the relationships between key variables and health-related outcomes. The findings support the proposed theoretical framework of AI technology use in the aging population and identify AI competency, an extended concept of perceived ease of use in TAM, as a significant factor that is positively associated with health-related life satisfaction. Overall, this study provides empirical evidence from a structural model linking AI competency, attitudes, and health-related life satisfaction among middle-aged and older adults using national survey data. The findings suggest that AI competency, rather than the frequency of AI use alone, may be more closely associated with perceived physical and mental well-being in later life, offering potential implications for future interventions.

Previous studies have consistently reported that digital literacy or competency is associated with life satisfaction or health-related quality of life, often through the mediating role of self-efficacy.^40–43 These studies primarily focused on well-established and widely implemented information and communication technologies or digital devices. This study extends this evidence by suggesting that AI technologies—despite their rapid evolution and unpredictable trajectory—may exhibit similar structural relationships with health-related quality of life. This finding suggests that theoretical frameworks that have traditionally been applied to technology adoption may also be valid in explaining AI-related technology use behaviors in middle-aged and older adults.

Based on the findings from this and previous studies, the importance of enhancing AI competency and self-efficacy for healthcare use should be emphasized in aging populations. Although several AI competency frameworks have been developed for students and educators,^44,45 none have been specifically designed for the general population or individuals with specific health conditions. Previous AI intervention studies involving older adults have reported improvements in general quality of life; however, these effects were modest and mixed, and often not directly linked to health-related outcomes.⁴⁶ Future studies should examine the impact of AI use on healthcare-specific outcomes in middle-aged and older adults and develop AI competency frameworks tailored to disease management. Such efforts could support the development of AI-enabled healthcare interventions that are better tailored to future aging populations.

In addition to AI competency, AI attitude also emerged as a significant factor associated with AI healthcare use frequency in the model. Notably, perceived AI competency and perceived AI helpfulness were both significantly associated with AI attitude, whereas AI attitude was directly associated with actual AI healthcare use. These findings suggest that older adults’ engagement with AI-based healthcare services may be shaped not only by technical competency but also by broader psychological readiness and positive attitude toward AI technologies. Positive attitudes toward AI may reflect greater trust, acceptance, and willingness to engage with technology, whereas concerns regarding data privacy, inequities, loss of human control may discourage engagement.^47,48 These findings suggest that promoting AI healthcare engagement among older adults may require not only technical training but also strategies that enhance positive attitude toward technology.

One notable finding of this study is that the frequency of AI healthcare use was not directly associated with health-related life satisfaction. This may be because AI competency reflects psychological characteristics, such as confidence and perceived control, which could directly contribute to mental well-being⁴⁹ regardless of actual engagement behaviors. This finding also suggests that simply using technology may not be sufficient to meaningfully influence perceived health-related well-being. Instead, feeling confident with AI and having a positive attitude toward it seems to have a bigger impact on the well-being of older adults.

However, their success in improving health outcomes seems to depend more on user confidence and understanding than just having access. This suggests that future nursing strategies should include training on both digital access and how to use specific AI features.^16,19,49 From a digital health perspective, these results show the need to design AI health technologies that focus on user understanding, confidence, and a sense of control. Programs that build AI literacy and skills may help older adults more than simply increasing technology use. These findings show that psychological and cognitive readiness, such as AI skills and attitudes, may matter more than how often older adults use these tools. For nursing, this means gerontological nurses should not only encourage access to AI-based services, but also help older adults build confidence, digital health skills, and a sense of control through targeted education and support. Future studies should examine how the quality of user–technology interaction moderates the relationship between AI competency and health-related outcomes over time.

Lastly, perceived AI healthcare helpfulness did not demonstrate a significant direct association with health-related life satisfaction. One possible explanation is that perceived helpfulness reflects evaluations of specific AI-related tasks or functions, whereas health-related life satisfaction represents a broader and more global assessment of physical and mental well-being. Consequently, perceived usefulness of individual AI healthcare services may not necessarily translate into higher overall life satisfaction among older adults.

South Korea is known for its advanced Internet infrastructure, and the sample characteristics in this study reflect this context in that all participants had Internet access and most had access to AI-enabled technologies, such as desktop computers or smartphones. Consequently, the findings may be most applicable to high-income countries where the majority of residents have reliable access to digital devices and the Internet. The participants in this study also had relatively favorable living conditions, as indicated by their levels of disability, education, and living status—only 12% lived alone. As unprivileged populations with lower education levels or those living alone may experience a greater digital divide and lower digital health literacy,²⁰ the current findings may have limited generalizability. Future research should include individuals with limited access to or awareness of AI healthcare technologies to better capture health-related impacts across diverse socioeconomic groups.

The study has several limitations that should be acknowledged. First, because all the data were cross-sectional and self-reported, causal relationships cannot be inferred. Some indicators also relied on a single item, potentially attenuating the true associations among the constructs. Several variables, including AI healthcare helpfulness, AI healthcare use frequency, and health-related life satisfaction, were measured using single-item indicators available in the national survey dataset. Although such measures are commonly used in large population-based surveys and have been associated with broader health and quality-of-life outcomes,^50,51 they do not permit assessment of internal consistency reliability and may provide limited evidence of construct validity for multidimensional constructs. Therefore, the findings should be interpreted cautiously.

Second, although this study aimed to examine the structural mechanisms of AI technology acceptance and its association with health-related outcomes, the measurement model showed borderline convergent and discriminant validity for the AI attitude construct despite its strong reliability. In older adults, perceived competence may dominate evaluative attitudes, leading to partial construct overlap. Also, this may reflect conceptual overlap with other variables in the model. Because more refined item development was not feasible in this study, given the nature of secondary analysis, future studies should incorporate revised and validated items to more precisely measure each construct in the model. Third, complete-case analysis was used to handle missing data, which may have introduced potential selection bias if excluded participants systematically differed from those included in the final analytic sample. Due to the cross-sectional design, causal relationships among the variables also cannot be inferred. The sample included only respondents who were aware of AI healthcare services, which could overrepresent individuals with greater digital access and socioeconomic resources, thereby limiting the generalizability of the findings.

Despite these limitations, the results revealed that access to laptops, tablet PCs, or smart accessories was associated with prior user experience with AI healthcare technologies. Although this study did not identify any direct effects of AI healthcare use frequency, previous evidence has shown positive associations between AI healthcare use and health outcomes, such as reduced hospitalization and fall rates.¹⁹ Therefore, improving access to AI-enabled digital devices among unprivileged populations and low-resource countries may create more opportunities for AI healthcare engagement, and ultimately contribute to better health outcomes across different countries and regions.

Conclusion

This study suggests that AI competency may be associated with health-related life satisfaction among middle-aged and older adults. These findings may provide useful considerations for the development of inclusive and user-centered AI healthcare services for aging populations. Future aging societies may require not only wider access to AI healthcare technologies but also support for older adults to develop the competency needed to engage with such technologies. Strengthening AI competency across diverse socioeconomic groups may also be an may help address disparities in digital health engagement among aging population.

Footnotes

Acknowledgements

During the preparation of this manuscript, an artificial intelligence tool was used for language editing, translation, and refinement of grammar. All study design, data analysis, interpretation of the results, and final manuscript content were conducted and verified solely by the authors.

ORCID iDs

Min Jung Kim

Hye Jin Chong

Author contributions

MJK, Conceptualization, Formal analysis, Methodology, Writing-review & editing, Writing-original draft, Funding acquisition; HJC, Methodology, Writing-review & editing, Writing-original draft.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (RS-2023-00250911). The funding body had no role in the design of the study, collection, analysis, and interpretation of data; or writing the article.

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 datasets generated or analyzed during this study are available on the public website.*

References

United Nations . World social report 2023: leaving no one behind in an ageing world, 2023, p. 162.

Yang

Lee

J-H

Lee

. Factors influencing health care technology acceptance in older adults based on the technology acceptance model and the unified theory of acceptance and use of technology: meta-analysis. Journal of Medical Internet Research 2025; 27: e65269. https://doi.org/10.2196/65269

Sulandari

Coats

Miller

, et al. A systematic review and meta-analysis of the association between physical capability, social support, loneliness, depression, anxiety, and life satisfaction in older adults. The Gerontologist 2024; 64: 1. gnae128. https://doi.org/10.1093/geront/gnae128

Bertolazzi

Quaglia

Bongelli

. Barriers and facilitators to health technology adoption by older adults with chronic diseases: an integrative systematic review. BMC Public Health 2024; 24: 506. https://doi.org/10.1186/s12889-024-18036-5

Liao

Zhou

. The mediating role of psychological resilience between social participation and life satisfaction among older adults in China. BMC Geriatrics 2022; 22: 948. https://doi.org/10.1186/s12877-022-03635-x

Lopez

Perez-Rojo

Noriega

, et al. Quality-of-life in older adults: its association with emotional distress and psychological wellbeing. BMC Geriatrics 2024; 24: 815. https://doi.org/10.1186/s12877-024-05401-7

Wang

, et al. Relationships among mental health, social capital and life satisfaction in rural senior older adults: a structural equation model. BMC Geriatrics 2022; 22: 73. https://doi.org/10.1186/s12877-022-02761-w

Kim

Hong

Choi

. Effects of serious games on depression in older adults: systematic review and meta-analysis of randomized controlled trials. Journal of Medical Internet Research 2022; 24: e37753. https://doi.org/10.2196/37753

Chang

C-H

Wei

C-C

Lien

W-C

, et al. The usability and effect of a novel intelligent rehabilitation exergame system on quality of life in frail older adults: prospective cohort study. JMIR Serious Games 2025; 13: e50669. https://doi.org/10.2196/50669

10.

Van Damme-Ostapowicz

Cybulski

Galczyk

, et al. Life satisfaction and depressive symptoms of mentally active older adults in Poland: a cross-sectional study. BMC Geriatrics 2021; 21: 466. https://doi.org/10.1186/s12877-021-02405-5

11.

Liu

Cai

Wittenberg

, et al. The Relationships Between Subjective Well-Being and Frailty: Staying With a Positive Mind, Stepping Away From Accelerated Aging. The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences 2025; 80: glaf001. https://doi.org/10.1093/gerona/glaf001

12.

Gutiérrez

Cabello

SanMartín

, et al. An mHealth App (NeoMayor) to Promote Healthy Lifestyles and Brain Health in Older Adults: Design and Validation Study. JMIR Aging 2025; 8: e71936. https://doi.org/10.2196/71936

13.

Chen

Ding

Wang

. Digital health for aging populations. Nature Medicine 2023; 29: 1623–1630. https://doi.org/10.1038/s41591-023-02391-8

14.

Hackstadt

Elasy

Gangaputra

, et al. Effects of a Patient Portal Intervention to Address Diabetes Care Gaps: Protocol for a Pragmatic Randomized Controlled Trial. JMIR Research Protocols 2024; 13: e56123. https://doi.org/10.2196/56123

15.

Chan

, et al. Health monitoring through wearable technologies for older adults: Smart wearables acceptance model. Applied Ergonomics 2019; 75: 162–169. https://doi.org/10.1016/j.apergo.2018.10.006

16.

Fernandes

Amoussou

da Rocha

, et al. Using Artificial Intelligence–Based Technologies for the Early Detection of Behavioral and Psychological Symptoms of Dementia: Scoping Review. JMIR Aging 2025; 8: e76074. https://doi.org/10.2196/76074

17.

Chien

S-C

Yen

C-M

Chang

Y-H

, et al. Use of artificial intelligence, internet of things, and edge intelligence in long-term care for older people: Comprehensive analysis through bibliometric, Google trends, and content analysis. Journal of Medical Internet Research 2025; 27: e56692. https://doi.org/10.2196/56692

18.

Frishammar

Essén

Bergström

, et al. Digital health platforms for the elderly? Key adoption and usage barriers and ways to address them. Technological Forecasting and Social Change 2023; 189: 122319. https://doi.org/10.1016/j.techfore.2023.122319

19.

Wilmink

Dupey

Alkire

, et al. Artificial intelligence–powered digital health platform and wearable devices improve outcomes for older adults in assisted living communities: Pilot intervention study. JMIR Aging 2020; 3: e19554. https://doi.org/10.2196/19554

20.

Yoon

Jang

Vaughan

, et al. Older adults’ internet use for health information: digital divide by race/ethnicity and socioeconomic status. Journal of Applied Gerontology 2020; 39: 105–110. https://doi.org/10.1177/0733464818770772

21.

Schroeder

Dodds

Georgiou

, et al. Older adults and new technology: Mapping review of the factors associated with older adults’ intention to adopt digital technologies. JMIR Aging 2023; 6: e44564. https://doi.org/10.2196/44564

22.

Tan

S-H

Yap

Y-Y

Tan

S-K

, et al. Determinants of Telehealth Adoption Among Older Adults: Cross-Sectional Survey Study. JMIR Aging 2025; 8: e60936. https://doi.org/10.2196/60936

23.

Holden

Karsh

B-T

. The technology acceptance model: its past and its future in health care. Journal of Biomedical Informatics 2010; 43: 159–172. https://doi.org/10.1016/j.jbi.2009.07.002

24.

Cimperman

Brenčič

Trkman

. Analyzing older users’ home telehealth services acceptance behavior—applying an Extended UTAUT model. International Journal of Medical Informatics 2016; 90: 22–31. https://doi.org/10.1016/j.ijmedinf.2016.03.002

25.

Zhang

Han

Dang

, et al. User acceptance of mobile health services from users’ perspectives: The role of self-efficacy and response-efficacy in technology acceptance. Informatics for Health and Social Care 2017; 42: 194–206. https://doi.org/10.1080/17538157.2016.1200053

26.

Lee

Coughlin

. PERSPECTIVE: Older adults' adoption of technology: An integrated approach to identifying determinants and barriers. Journal of Product Innovation Management 2015; 32: 747–759. https://doi.org/10.1111/jpim.12176

27.

Forsman

Nordmyr

Matosevic

, et al. Promoting mental wellbeing among older people: technology-based interventions. Health Promotion International 2018; 33: 1042–1054. https://doi.org/10.1093/heapro/dax047

28.

Aggarwal

Xiong

Schroeder-Butterfill

. Impact of the use of the internet on quality of life in older adults: review of literature. Primary Health Care Research & Development 2020; 21: e55. https://doi.org/10.1017/S1463423620000584

29.

National Information Society Agency . The Report on the Digital Divide. , 2024.

30.

Davis

. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 1989; 13: 319–340. https://doi.org/10.2307/249008

31.

Zhu

Wan

, et al. Older adults’ self-perception, technology anxiety, and intention to use digital public services. BMC Public Health 2024; 24: 3533. https://doi.org/10.1186/s12889-024-21088-2

32.

Baek

E-M

Choi

Kang

S-H

. Exploring the relationship between digital competency and life satisfaction Among the elderly. Digital Health 2025; 11: 20552076251358756. https://doi.org/10.1177/20552076251358756

33.

Xie

Fang

Wang

, et al. Providing digital technology training as a way to retain older workers: The importance of perceived usefulness and growth need. Work, Aging and Retirement 2023; 9: 376–392. https://doi.org/10.1093/workar/waad004

34.

Adnan

Irvine

Williams

, et al. Improving Acceptability of mHealth Apps—The Use of the Technology Acceptance Model to Assess the Acceptability of mHealth Apps: Systematic Review. Journal of Medical Internet Research 2025; 27: e66432. https://doi.org/10.2196/66432

35.

Gan

McGillivray

Han

, et al. Effect of engagement with digital interventions on mental health outcomes: a systematic review and meta-analysis. Frontiers in Digital Health 2021; 3: 764079. https://doi.org/10.3389/fdgth.2021.764079

36.

Anderson

Gerbing

. Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin 1988; 103: 411–423. https://doi.org/10.1037/0033-2909.103.3.411

37.

Kline

. Principles and practice of structural equation modeling. Guilford publications, 2023.

38.

Fornell

Larcker

. Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research 1981; 18: 39–50. https://doi.org/10.2307/3151312

39.

Ausserhofer

Piccoliori

Engl

, et al. Community-Dwelling Older Adults’ Readiness for Adopting Digital Health Technologies: Cross-Sectional Survey Study. JMIR Formative Research 2023; 8: e54120. https://doi.org/10.2196/54120

40.

Xie

. A 3-Wave Longitudinal Study of eHealth Literacy and Older People's Health-Related Quality of Life in China: The Mediating Role of General Self-Efficacy. Journal of the American Medical Directors Association 2024; 25: 105310. https://doi.org/10.1016/j.jamda.2024.105310

41.

Lozoya

SVM

Guirado

MAZ

González

, et al. Use of technologies and self-efficacy in older adults. IEEE Revista Iberoamericana de Tecnologias Del Aprendizaje 2022; 17: 125–130. https://doi.org/10.1109/rita.2022.3166870

42.

Lee

Suk

Kim

, et al. Do ICT Competence in and ICT Service Use Affect Life Satisfaction? Focusing on Mobile ICT Services. International Conference on Human-Computer Interaction. Springer, 2022, pp. 156–163.

43.

Hatamnezhad

Ghafari Ashtiyani

Seyedi

. Investigating the Relationship Between Electronic Literacy and Quality of Life of the Elderly in Arak, Iran. Bulletin of Science, Technology & Society 2021; 41: 3–9. https://doi.org/10.1177/02704676211007360

44.

Miao

Shiohira

. AI Competency Framework for Students. UNESCO Publishing, 2024.

45.

Cukurova

Miao

. AI Competency Framework for Teachers. UNESCO Publishing, 2024.

46.

Seo

Jang

Seo

. Effect of AI intervention programs for older adults on the quality of life: A systematic review and meta-analysis of randomized controlled trials. Digital Health 2025; 11: 20552076251324014. https://doi.org/10.1177/20552076251324014

47.

Chen

Aquino

YSJ

, et al. Multi-stakeholder preferences for the use of artificial intelligence in healthcare: A systematic review and thematic analysis. Social Science & Medicine 2023; 338: 116357. https://doi.org/10.1016/j.socscimed.2023.116357

48.

Scott

Carter

Coiera

. Exploring stakeholder attitudes towards AI in clinical practice. BMJ Health & Care Informatics 2021; 28: e100450. https://doi.org/10.1136/bmjhci-2021-100450

49.

Naiseh

Babiker

Al-Shakhsi

, et al. Attitudes towards AI: The interplay of self-efficacy, well-being, and competency. Journal of Technology in Behavioral Science 2025; 10: 1–14. https://doi.org/10.1007/s41347-025-00486-2

50.

Bowling

. Just one question: If one question works, why ask several? BMJ Publishing Group Ltd, 2005, pp. 342–345.

51.

Cheung

Lucas

. Assessing the validity of single-item life satisfaction measures: Results from three large samples. Quality of Life Research 2014; 23: 2809–2818. https://doi.org/10.1007/s11136-014-0726-4