Designing Sustainable Digital Platforms for Ageing Societies: A User-Centred Multi-Level Theoretical Framework

Abstract

With the intensification of population ageing and the increasingly diverse service needs of older adults, existing digital elderly care platforms generally exhibit fragmentation in functional integration, understanding of needs, and service coordination, making it difficult to effectively respond to the complex challenges faced by urban ageing populations. To fill this gap, this study starts from a service design perspective and adopts Constructivist Grounded Theory (CGT) to construct a theoretical model, proposing a three-tier framework that encompasses seven core user needs, four platform response mechanisms, and three categories of service outcomes. A questionnaire survey was subsequently conducted in the Pearl River Delta region of China, collecting 352 responses, of which 322 were valid. Through Exploratory Factor Analysis (EFA), correlation analysis, and multiple regression analysis, the structural stability and predictive validity of the proposed “User Needs-Platform Mechanisms-Service Outcomes” (UN-PM-SO) model were verified. The research results confirm that the theoretical model constructed in this study has good logical consistency and empirical support. Based on this model, a series of concrete design framework recommendations are further proposed, aiming to guide the sustainable and inclusive development of future smart elderly care platforms. The findings of this study not only respond to the urgent global demand for age-friendly digital infrastructure but also demonstrate the sustainable value of smart elderly care platform design in terms of social inclusion, resource efficiency, and environmental friendliness, providing a feasible and theory-based design logic and governance pathway for promoting social sustainability.

1. Introduction

With the acceleration of global population ageing, how to respond to the care challenges of ageing societies through digital technology has become a common policy and service issue across countries. According to statistics from the World Health Organisation, by 2050, the global population aged 60 and over will exceed 2 billion, and the needs of older adults in areas such as health management, daily support, and social connection are becoming increasingly diverse and context-specific [1]. In recent years, many countries—including OECD member states—have actively introduced digital platforms as tools for integrating public care resources and coordinating service interactions, aiming to overcome the long-standing fragmentation and delayed responsiveness of traditional care systems. However, in practical implementation, many platforms still face challenges such as insufficient understanding of user needs, static module design, and weak cross-sector collaboration. These issues result in poor accessibility, low user participation, and heightened feelings of exclusion among older adults in digital service environments [2].

Against this global backdrop, China is rapidly entering a phase of deep population ageing, with a substantial increase in the number of older adults, posing multi-layered, context-specific, and dynamically evolving challenges to community-based elderly care services. According to data from the Seventh National Population Census, by the end of 2020, the population aged 60 and above had reached 264 million, accounting for 18.7% of the total population [3]. In response to similar demographic trends, some ageing societies have begun to restructure their digital elderly care platforms through demand-oriented and scenario-integrated approaches. For instance, Japan has established a digital health management loop characterized by “consultation–coordination–sharing” by integrating telemedicine with community pharmacists and family caregivers, thereby enhancing care efficiency and user engagement [4]. The Netherlands has adopted decentralized governance and small-scale autonomy, supported by lightweight ICT systems and regional coaching mechanisms, to facilitate localized collaboration across multidisciplinary teams and to build a highly responsive community-based care model [5].

In comparison, although Chinese authorities have actively promoted smart elderly care and digital transformation in recent years—articulated in policy documents such as the Action Plan for the Development of Smart Health and Elderly Care Industry (2021–2025) and the 14th Five-Year Plan for the Development of National Ageing Undertakings and Elderly Care Service Systems [6,7]—the current system remains fragmented in terms of resource allocation and service models. Most existing platforms are still based on the simple accumulation of functional modules, lacking a holistic design grounded in user-centred logic [8,9]. The disconnection between platform design and real-life elderly scenarios, coupled with operational complexity and unfriendly interfaces, limits both accessibility and participation, further exacerbating the digital divide [10,11,12]. Therefore, to gain an in-depth understanding of the structural dilemmas in the design and practice of smart elderly care platforms in China, it is no longer sufficient to remain at the level of describing “technological access”; instead, it is necessary to situate the issue within the multi-level framework of the digital divide.

Within the multi-level framework of the digital divide, the first level concerns access, referring to the availability of networks and devices; the second level pertains to skills and usage, involving operational competence and application contexts; the third level emphasizes outcomes/benefits, focusing on whether usage leads to tangible effects such as health management, social connection, and everyday convenience [13,14]. Some studies further extend to a “fourth level,” highlighting the constraints on sustained participation and the risks of digital disengagement [15]. Cross-national data indicate that disparities at the access level have markedly diminished: by 2023, nearly 60% of the EU population aged 65–74 had used online banking or searched for health information [16]; in China, as of June 2024, the internet penetration rate among people aged 60 and above had reached 52.0%, with 99.7% accessing the internet via mobile phones, and over 70% of elderly netizens using mobile payments [17]. However, significant gaps remain in skills and benefits: in the EU, only about 30% of older adults possess at least basic digital skills, while in China, fewer than 20% consider themselves proficient in operating smartphones or online applications. Although statistical measures vary, they consistently point to the same trend—the contemporary “digital divide” is primarily manifested in insufficient skills and limited benefit conversion, rather than device shortages [18,19]. Accordingly, this study focuses on the second level (skills and usage) and the third level (benefit disparities) in constructing and validating the model and explores how platform design can transform low-frequency participation into stable benefits to address the core challenges of smart elderly care. It should also be emphasized that low-frequency or selective non-use does not necessarily imply passivity or exclusion; some older adults may deliberately reduce usage due to risk preferences or privacy concerns, which still requires design and contextual support to be translated into sustainable benefits.

As one of the most economically dynamic and highly urbanized regions in China, the Pearl River Delta (PRD) exhibits three typical structural challenges in addressing ageing and digital transformation. First, the elderly population in the region is highly diverse, comprising local residents, returning retirees, and cross-regional accompanying caregivers, resulting in heterogeneous demand patterns [20]. Second, elderly care services span multiple departments—including health care, civil affairs, and community management—but lack clearly defined responsibilities and horizontal coordination, resulting in service discontinuities [21]. Third, there is a pronounced disparity in resource allocation across cities within the region, leading to significant differences in service accessibility and adaptability [22]. At the same time, the PRD also serves as a high-density policy pilot zone under China’s “experiment-driven governance” model, where numerous cities have initiated reforms in integrated medical-care services and smart elderly care, showcasing both institutional innovation and implementation gaps [23]. This study selects six representative cities across different metropolitan clusters and functional positions to serve as empirical sites for observing the integration potential of digital platforms under conditions of high-demand heterogeneity and fragmented governance.

Accordingly, this research addresses the above challenges through the lens of service design. It adopts Constructivist Grounded Theory (CGT) to conduct qualitative interviews and category construction, followed by a quantitative survey for model verification. The objective is to explore a theoretical framework that integrates the logic of need generation with platform design and to further verify its applicability and operability in promoting equity of access, resource efficiency, and long-term sustainability in smart elderly care services. These issues are also closely aligned with the United Nations Sustainable Development Goals (SDGs), particularly SDG 3 (Good Health and Well-being), SDG 10 (Reduced Inequalities), and SDG 11 (Sustainable Cities and Communities), which constitute both the empirical and theoretical motivation of this study. The aim is to develop a theoretical framework that integrates need generation and platform design logic and to further assess its applicability and operational viability. Prior studies have shown that platforms, as a form of digital infrastructure, hold the potential to integrate service modules, behavioral data, and stakeholder interactions [24,25]. However, theoretical models centred explicitly on the logic of elderly users’ needs remain scarce.

Based on the above research objectives and methods, the structure of this paper is arranged as follows: Section 2 reviews two main strands of the literature—health-oriented needs of older adults and applications of digital elderly care platforms—and identifies research gaps and theoretical motivations. Section 3 explains the research methodology and process, including semi-structured interviews, construction of user journey maps, the four-stage coding procedure of Constructivist Grounded Theory (CGT), and the derivation of a questionnaire from qualitative categories, along with sampling and statistical methods. Section 4 presents the research results, covering the three-tier “User Needs–Platform Mechanisms–Service Outcomes (UN-PM-SO)” model generated from CGT coding, the test of theoretical saturation, and quantitative validation (reliability, validity, and EFA, as well as correlation and multiple regression analyses). Section 5 provides the discussion, engaging in dialogue with theories such as TAM, UTAUT, HCI, and recommender systems, and it translates the model into platform design strategies and governance implications. Section 6 concludes the study by summarizing the findings and contributions while also acknowledging research limitations and outlining directions for future research.

2. Literature Review

To clarify the core challenges and demand logic underlying the design of smart elderly care platforms, this study focuses on two primary bodies of literature: (1) health-oriented needs of older adults and (2) the design and application of digital platforms for elderly care. These two domains serve as the analytical foundation for the construction of the theoretical model and its translation into service design strategies.

2.1. Review of Health-Oriented Needs Among Older Adults

As the ageing process accelerates, the needs of older adults have evolved beyond basic physiological care into a dynamic structure encompassing health, emotional, and social dimensions. Most studies regard health as the core driver of need formation, noting that health deterioration is often accompanied by increased loneliness, emotional distress, and social withdrawal, which in turn trigger a chain of additional needs such as transportation assistance, psychological support, and functional compensation [26,27]. These findings suggest that health conditions not only shape the physical behavior patterns of older adults but also profoundly influence their life satisfaction and willingness to engage socially.

On a psychological level, existing research indicates that psychological well-being is typically built upon the interplay between physical health, emotional regulation, and social connectivity [28]. In the absence of emotional support and positive interactions, it becomes more difficult for older adults to maintain functional capacity and psychological stability, resulting in an expanded and compounded structure of needs. Beyond internal factors, mobility—particularly transportation—has been identified as a crucial external condition affecting the fulfillment of health needs. Gimie et al. reported that more than 3 million older adults in the United States rely on public transportation for medical visits, highlighting its direct linkage to health behaviors [29]. Zhang et al. found that perceived transportation inclusiveness significantly enhances subjective well-being [30], while Maresova et al. further emphasized that mobility is jointly constrained by physiological and psychological factors, and that limitations in this domain can weaken social participation and emotional connection, thereby adversely affecting health [31]. Simultaneously, socio-cultural contexts also exert a key moderating effect on the formation and reconstruction of needs. For example, Gadermann et al. demonstrated that life stress events and shifts in social support networks significantly impact the mental state of older adults and recalibrate their needs for intimacy and social interaction, highlighting the contextual and situational variability of need structures [32].

Taken together, the formation and evolution of elderly needs are shaped by an interplay of physiological, psychological, social, and cultural factors. This complex structure is characterized by dynamic generation, chain expansion, and contextual reconfiguration [33]. However, the current literature still presents three key limitations in addressing the dynamic and contextual nature of such needs: (1) most studies rely on static classifications and unidirectional analyses, lacking a holistic depiction of interwoven needs; (2) there is insufficient theoretical exploration of how health conditions initiate the generation of other needs; and (3) the processes through which older adults experience the transformation and migration of needs—especially how these are identified and addressed within digital interfaces—remain underexplored. Therefore, future research should emphasize the logical interconnections among needs, incorporate lived contexts and cultural variables, and develop theoretical models with generative, transformative, and contextual dimensions to better support scenario-based smart elderly care services.

2.2. Review of the Design and Application of Digital Platforms for Elderly Care

Contemporary digital platforms for elderly care can be broadly categorized into three typical models: government-led, enterprise-led, and community-led platforms. Government-led platforms focus on integrating healthcare, caregiving, and community governance resources, aiming to enhance public service coverage and governance efficiency through multi-agency collaboration [34,35]. Enterprise-led platforms are more market-driven, emphasizing user experience and the application of technologies such as cloud computing, artificial intelligence, and data analytics in health monitoring and daily assistance [36,37]. Meanwhile, community-led platforms have shown increasing potential in recent years, promoting the integration of home- and community-based services through community nodes to establish participatory care networks involving multiple stakeholders [38,39]. Although all three types of platforms take health management as a core function, studies have highlighted a lack of interoperability standards and integrative coordination mechanisms, leading to resource fragmentation and inadequate adaptability [40,41].

At present, most platform designs remain centred on technological modules, focusing on validating the performance of IoT systems, wearable devices, and telecare solutions but rarely addressing the logic of need transformation across diverse life scenarios [42,43,44]. The services offered by these platforms are often constrained by predefined scenarios and static modules, lacking the capacity for dynamic adaptation to real-world living conditions [45]. In response, some studies have shifted toward building intelligent platforms that incorporate behavioral sensing, health prediction, and personalised recommendation strategies [46,47,48], emphasizing the need for dynamic adaptation and cross-module integration [49,50,51].

Building on this, the service design perspective has emerged as a critical supplement to platform construction. Relevant studies have shown that service design—through user behavior insights, co-design processes, and system flow optimisation—can effectively integrate cross-module functions and accommodate the participation of diverse actors [52,53]. For example, He Xuemei et al. employed service blueprints and user behavior chain analysis to develop platform restructuring strategies, highlighting the integrative role of service design in multi-touchpoint and multi-role scenarios [54]. These studies offer valuable theoretical support and strategic references for enhancing the agility, sustainability, and inclusiveness of smart elderly care platforms in complex service environments.

2.3. Research Gaps and Theoretical Motivation of This Study

Although research on ageing-oriented digital platforms has gained increasing attention, the existing literature still suffers from a lack of coherent theoretical frameworks and fragmented viewpoints. A considerable number of studies have adopted general technology acceptance models, such as the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT), to explore older adults’ behavior in using digital technologies. However, these models are often applied in a structurally simplified manner, overlooking the contextual, emotional, and behavioral complexities that older adults encounter in real-world community settings. As a result, they fall short in accurately predicting actual usage intentions and behaviors [55,56,57]. Furthermore, while some studies have focused on individual platform case studies, they generally lack transferable theoretical models capable of explaining the interactive logic between user needs and service mechanisms [58]. This reflects a broader limitation in the field: the absence of integrative, empirically grounded frameworks that connect user demand generation, service design, and platform interaction in a systematic and dynamic way.

To address these research gaps, this study adopts Constructivist Grounded Theory (CGT) as its methodological foundation. By centering on the lived experiences and perspectives of older adults and community stakeholders, the study seeks to construct a multi-layered theoretical model that captures the dynamic interplay between user needs, service mechanisms, and platform outcomes. On this basis, the study further frames “equity of access, resource efficiency, and long-term operation” as its key analytical perspective. This bottom-up, user-centred approach not only compensates for the limitations of prior theoretical models but also offers a practical reference for the sustainable design of digital platforms in ageing societies.

3. Research Methodology

To gain an in-depth understanding of the multi-level service needs encountered by older adults in community-based elderly care settings, and to develop a theoretical framework that supports digital platform design, this study adopts a mixed-methods approach combining qualitative and quantitative methods. Anchored in a service design perspective, the research emphasizes the identification of need-generation mechanisms and potential points of platform intervention based on users’ actual behaviors and service interaction contexts. The overall methodological process comprises three key stages: (1) semi-structured interviews and qualitative data analysis; (2) a theory-building process based on CGT; and (3) quantitative survey design derived from theoretical categories, followed by statistical validation (see Figure 1).

3.1. Semi-Structured Interviews and Qualitative Data Collection

3.1.1. Interview Design and Theoretical Foundations

To enhance the reliability and validity of the interview questions, the design process drew upon the existing literature that categorizes types of needs and characteristics of service interactions among older adults. For health management scenarios, the study referenced Farage et al.’s work on multi-layered needs related to health monitoring and self-management among older adults [59]. In addressing platform usability, findings from Fang et al. on digital health platform usability and operational barriers were incorporated [11]. Questions related to emotional support were informed by studies from Upasen and Gadermann, which emphasized the behavioral consequences of loneliness and emotional health, including the need for social networks and sources of support [28,32]. For rehabilitation and care scenarios, the work of Maresova and J.-J. Chen on chronic illness management and long-term care challenges guided the formulation of interview strategies around individualized care experiences [27,31]. Platform functionality and integration needs were explored through insights from Sharma and Nasr, who discussed discrepancies between interface design and the behavioral expectations of older users, particularly in terms of ease of use, multifunctional integration, and adaptive optimization [44,45].

Interview questions were centred around six key service scenarios: health management, rehabilitation and care, emotional support and social interaction, daily life assistance, community service use, and personalised platform design. Each scenario featured open-ended questions targeting key service touchpoints and potential barriers. The scenario categories were contextually constructed following service design principles and further validated through the literature [47]. During the interviews, no predefined classification of needs was imposed. Instead, the study employed a CGT perspective to allow for the natural emergence of data, which were subsequently categorized into analytical domains (see

Table 1. Scenario Classification and Corresponding Interview Outline Design.

Service Scenario	Scenario Description	Interview Questions
A. Health Management Services	Focuses on health checkups, medication, and health monitoring, as well as related service access and barriers.	A-1. How do you usually manage your daily health?
		A-2. Are the current community healthcare services meeting your needs?
B. Rehabilitation and Long-term Care Services	Addresses challenges in chronic disease management, post-treatment rehabilitation, and long-term care support.	B-1. Have you ever received rehabilitation or long-term care services? What problems did you encounter?
		B-2. In your opinion, what are the current shortcomings of rehabilitation services?
C. Emotional Support and Social Interaction Services	Explores the impact of insufficient emotional support or limited social interaction on health status and service participation.	C-1. When feeling unwell, do you also feel lonely?
		C-2. Are you able to obtain sufficient emotional support from your surroundings?
D. Integrated Daily Life Support Services	Focuses on difficulties related to transportation, household support, and daily living assistance.	D-1. What are your basic non-medical needs in daily life?
		D-2. Are these needs currently well supported?
E. Community Service Access Services	Examines the accessibility and usability of community service systems.	E-1. Have you encountered any difficulties when using community services?
		E-2. Does the service system design meet your expectations?
F. Personalised Service Design Services	Investigates the feasibility of providing personalised services tailored to individual needs.	F-1. If a unified platform could provide all of the above functions, what features would you most desire?
		F-2. What type of interface design would be easiest for you to use?

). This design of the interview guide not only addressed functional and health-related aspects but also explicitly incorporated socio-emotional domains, consistent with design case research highlighting older adults’ social and emotional needs as central to product and service development [60].

3.1.2. Interview Sample and Data Sources

The study sample consisted of older adults and frontline community workers from six cities in the PRD region of China (Guangzhou, Shenzhen, Foshan, Dongguan, Zhuhai, and Zhongshan), reflecting diverse local conditions in elderly care policy implementation and social structures. Inclusion criteria for older adult participants were: age 60 or above, basic self-care and digital device operation abilities, and the ability to articulate experiences related to health management or service usage. Individuals with severe cognitive impairment or communication difficulties were excluded. To supplement perspectives, the study also included professional social workers from community service centers as representatives of service providers, offering third-party interpretations regarding policy implementation and resource gaps [61]. All participants signed informed consent forms and explicitly agreed to the anonymous academic use of their data. The study adhered to basic research ethics principles and received approval from the Human Research Ethics Committee. The study followed basic ethical principles, and all respondents voluntarily participated with full awareness of the research purpose. All personal information was de-identified to ensure privacy. Basic demographic information (e.g., age, gender, city of residence, and pre-retirement occupation) is provided in

Table 2. Basic Information of Interviewees.

Participant Group	Age	Gender	Education Level	Pre-retirement Occupation	Region
Community elderly	63	Female	Junior high school	Textile factory worker	Dongguan
	62	Male	Senior high school	Laborer
	63	Male	Vocational school	Construction worker
	70	Female	Technical secondary school	Primary school teacher	Foshan
	70	Female	Primary school	Laborer
	63	Female	Junior college	Market vendor
	72	Female	Senior high school	Community care worker
	65	Male	Vocational school	Technician	Guangzhou
	68	Male	University	University lecturer
	64	Male	University	Medical administration
	65	Female	Primary school	Laborer	Shenzhen
	70	Female	Junior high school	Bus driver
	69	Male	Junior college	Laborer
	68	Female	Junior high school	Housewife
	67	Male	Senior high school	Taxi driver
	63	Male	Senior high school	Retail salesperson	Zhuhai
	65	Male	Primary school	Tailor
	67	Female	Senior high school	Manufacturing worker
	66	Female	Senior high school	Primary school teacher	Zhongshan
	66	Female	Junior high school	Primary school teacher
	60	Female	Primary school	Owner of a sewing shop
	62	Male	Junior high school	Secondary school teacher
Community worker	Undisclosed	Female	Bachelor	/	Foshan
	42	Male	Bachelor	/	Guangzhou
	28	Female	Master	/	Shenzhen
	35	Female	Bachelor	/	Dongguan
	49	Male	Senior high school	/	Zhongshan

.

3.1.3. User Journey Mapping

In service design-oriented qualitative research, user journey maps serve as essential tools to structure user experiences and visualize service interaction flows. These maps help identify pain points, need triggers, and potential intervention opportunities for digital platforms. Based on the six service scenarios, this study extracted contact points, perceived needs, usage barriers, and service expectations from the semi-structured interview data and constructed a cross-scenario user journey map (see Figure 2).

Unlike top–down models that deduce needs based on pre-existing categories, this journey map was directly constructed from participants’ narratives. Through semantic matching and contextual reconstruction, the map illustrates elderly users’ lived experiences of service interaction within real-life environments. This step was conducted prior to grounded theory coding to provide a contextually rich experiential foundation that supports subsequent need extraction and institutional gap analysis.

The journey map revealed three core challenges in community-based elderly care services:

(1)

Lack of service integration: Participants commonly reported discontinuities between health and rehabilitation services, a shortage of emotional and social support resources, and weak horizontal coordination between daily living and medical services.

(2)

Misalignment between demand and service supply: Most services offered were standardized and failed to account for individual health conditions and life contexts, lacking both flexibility and adaptability.

(3)

Accessibility and resource bottlenecks: Barriers such as inadequate equipment and staffing, inflexible social work arrangements, and age-unfriendly transportation and appointment systems reduced both willingness and ability to engage with services.

In terms of service expectations, older adults cared not only about whether services were offered but also about whether they were conveniently accessible, sustainable, and personally tailored. Most participants supported an integrated service design approach and hoped for a unified process that includes health, daily life, transportation, and emotional support, with dynamic adjustments based on changing needs. The journey map not only provided concrete situational foundations for subsequent grounded theory coding but also functioned as a contextual mediator linking user perspectives with platform design logic, laying the groundwork for cross-scenario integration strategies.

3.2. Application of Grounded Theory and Qualitative Analysis

Grounded Theory, introduced by Glaser and Strauss in 1967, is a qualitative research methodology that emphasizes building theory directly from data rather than verifying pre-existing theoretical frameworks. This method is particularly well-suited for exploring emerging social phenomena and is widely regarded as one of the most rigorous approaches to qualitative research [62,63]. With the global trend of population ageing, the needs of older adults are becoming increasingly diverse and dynamic, rendering traditional static demand models inadequate for addressing the interwoven and evolving nature of such needs. Therefore, this study adopts CGT to explore the dynamic characteristics of elderly needs in greater depth.

CGT emphasizes the identification of needs and pain points as an evolving process that adapts to changing contexts and sociocultural environments. This perspective enables the generation of adaptive theories in response to rapidly shifting social realities. Compared with Classic Grounded Theory, CGT is more effective in capturing the changing needs of older adults within specific sociocultural contexts. Particularly in the field of digital service platform design, CGT supports design flexibility by enabling the construction of theoretical frameworks directly from data that reflect the real needs of elderly users. Thus, adopting CGT in this study not only addresses the limitations of traditional need-based models but also supports the generation of responsive and adaptive solutions in digital design.

Through the analysis of elderly users’ service needs and pain points in community settings, the study finds that existing community services are often fragmented and slow to respond, leaving health management, emotional support, and social interaction needs largely unmet. While community-based service models remain the mainstream approach, their limitations have prompted this study to explore digital platform design as a potential solution. Digital platforms offer the capacity to integrate services, deliver real-time responses, and enable personalised design based on user needs [64]. Accordingly, as the grounded theory analysis progressed, the study gradually shifted its analytical focus toward digital platform design. This shift, grounded in the identification of service inadequacies, revealed how platforms can effectively address multi-dimensional elderly needs. Without deviating from the original research goals, this pivot further clarified the potential of digital platforms in overcoming the limitations of fragmented community services.

This study is organized around three guiding questions: (1) What are the major pain points in current community services, and which elderly needs remain unmet? (2) What mechanisms can digital platforms employ to respond to these needs? (3) How can digital platforms better address elderly users’ needs and improve quality of life? These questions served as the foundation for the coding process, allowing the study to systematically explore how digital platforms can be designed to effectively meet elderly users’ needs.

3.3. Coding Procedures and Analytical Objectives

The data analysis in this study followed the four stages of CGT: initial coding, focused coding, axial coding, and theoretical coding [65]. Each stage was designed to extract key concepts from the data and map them to service pain points, user needs, and potential intervention mechanisms. The analysis was conducted using Microsoft Excel for both initial and focused coding. The specific coding steps and corresponding objectives are detailed in

Table 3. CGT Coding Steps and Objectives.

Coding Stage	Objective of the Stage
A. Initial Coding	Initial coding involves breaking down interview transcripts line-by-line to identify meaningful units, allowing detailed interpretation of each participant’s narrative. The goal is to capture older adults’ service pain points, needs, and emotional responses while maintaining semantic consistency. Key pain points and service demands are inductively identified during this phase.
B. Focused Coding	After initial coding, this stage involves consolidating and abstracting frequently occurring or conceptually representative codes to extract core subthemes. Through constant comparison, these subthemes enable preliminary categorization of service demands, providing the basis for later theoretical construction.
C. Axial Coding	Axial coding integrates the concepts generated from focused coding, analyzing their causal relationships. This phase constructs a “Condition-Action-Result (C-A-R)” model, aiming to reveal the underlying mechanisms behind various types of service needs and how they interact with platform features.
D. Theoretical Coding	In the theoretical coding phase, prior findings are synthesized to generate abstract yet operable theoretical models. The outcome is a logic-based, generalizable framework that serves as the conceptual foundation for platform design and theoretical integration.

.

3.4. Quantitative Survey for Model Validation

Following the qualitative analysis based on CGT, this study employed a quantitative survey to validate the theoretical framework developed during the earlier phase and to examine the relationships between categories, as well as their impact on the willingness of older adults to use smart community elderly care platforms. The survey method, as a structured approach to collecting respondents’ views, attitudes, and behaviors, is particularly suitable for large sample sizes and enables statistical validation and inference of underlying patterns.

3.4.1. Questionnaire Module Design

The questionnaire was developed based on the core and subcategories extracted from the qualitative phase, transforming abstract qualitative concepts into measurable quantitative indicators. This step serves as a crucial bridge between theoretical construction and empirical validation [66]. The items were designed to reflect key concepts distilled from the qualitative findings, ensuring alignment between qualitative insights and the quantitative tool [67]. A five-point Likert scale was adopted (1 = strongly disagree, 5 = strongly agree) to ensure statistical tractability. To enhance respondent comprehension, all questionnaire items were phrased using clear and concise language, avoiding technical jargon and complex sentence structures.

3.4.2. Questionnaire Distribution Strategy

The questionnaire was distributed via a combination of online and paper-based formats to accommodate older adults with varying levels of digital literacy. Respondents familiar with digital technologies could complete the questionnaire online, while those less proficient were given paper copies, with assistance provided by the research team or community social workers. This approach ensured broad data coverage and inclusivity across different levels of digital competency.

Participants were aged 60 and above and drawn from cities in the PRD region, including Guangzhou, Shenzhen, and Dongguan. These areas differ significantly in their social, economic, and elderly care service configurations, allowing the study to capture how diverse socio-economic conditions shape demand and platform usage behaviors. The sample design thus ensured adequate representativeness across heterogeneous user groups and enabled the exploration of variations in health management, platform interaction, and service expectations.

3.4.3. Data Analysis Methods

After collection, survey data underwent cleaning and were subjected to the following quantitative analyses based on the study objectives:

1.

Exploratory Factor Analysis (EFA):

Used to examine the structural validity of the questionnaire and ensure the internal consistency of items within each theoretical category, thus testing the reliability and construct validity of the measurement model.

2.

Correlation Analysis:

Conducted to explore relationships among the core categories and to preliminarily validate potential interactions and associations between need domains.

3.

Multiple Regression Analysis:

Applied to assess the influence of different categories on elderly users’ willingness to use digital platforms. This analysis helped establish causal pathways between service design features and need fulfillment, offering theoretical support for platform optimisation strategies.

4. Results

4.1. Results of Grounded Theory Analysis

This study applied the four-stage analytical strategy of CGT to conduct in-depth semantic coding of verbatim interview transcripts from 28 participants. The process included initial coding, focused coding, axial coding, and theoretical coding. The final result was a three-tier theoretical model structured around the core logic of “UN-PM-SO model”. The analysis was guided by three central questions: (1) What are the gaps and pain points in existing community services for older adults? (2) How can digital platforms address and translate these unmet needs? (3) What behavioral or experiential outcomes may result from platform use? Due to the large volume of Chinese-language interview data, only a selection of coded results is presented below.

To enhance traceability and semantic consistency, a hierarchical coding system was established. “T” and “S” denote elderly participants and community workers, respectively. Original utterances were labeled as “T/S + participant ID—sentence number” (e.g., T4-21). Initial codes were annotated as “a” (e.g., T4-a21), and focused codes were further marked as “aa” (e.g., T4-aa21). These were then abstracted into subcategories (e.g., A1, B2) and subsequently grouped into nine major categories (e.g., AA1, BB1), forming the conceptual basis of the “Needs—Mechanisms—Outcomes” three-tier model. Theoretical coding then established logical relationships and contextual linkages among categories.

4.1.1. Initial Coding Phase

In the initial coding phase, the interview transcripts were deconstructed line by line to extract semantic units, resulting in a total of 1650 initial codes. These codes reflected concrete needs and behavioral patterns related to health management, daily living, social interaction, platform functionality, and institutional policy. They also documented users’ disappointment with current community services and expectations of digital platforms.

Three major thematic clusters emerged: (1) Fragmentation of community services, resource scarcity, and delayed responses, which were coded as “insufficient health management support,” “lack of social support,” etc.; (2) Expectations for digital platform functions, such as “health monitoring tools,” “voice-interaction interfaces,” and “personalised reminders”; (3) Anticipated positive outcomes from platform usage, such as “improved health management,” “enhanced emotional support,” and “better quality of life.” These initial codes provided grounded observations of elderly users’ real-life contexts and laid a solid foundation for semantic abstraction and category development (See

Table 4. Excerpt of Selected Initial Codes.

Code ID	Original Transcript (in Chinese)	Segment Code	Initial Coding (in Chinese)
T4-21	“I still hope to receive continuous rehabilitation guidance … and ongoing rehabilitation training.”	T4-a21	Hopes for post-discharge rehabilitation guidance and structured training to enhance recovery effectiveness.
T6-26	“More rehabilitation therapists … to correct my movements.”	T6-a26	Expects more individualized instruction from therapists to improve rehabilitation outcomes.
T1-29	“The current rehabilitation and nursing services … are just passable.”	T1-a29	Perceives current rehabilitation and nursing services as substandard, with a need for quality improvement.

).

4.1.2. Focused Coding Phase

To enhance conceptual coherence and ensure category stability, a “focused concept” layer was established between initial codes and subcategories. Semantically related and functionally similar initial codes were integrated and grouped (see

Table 5. Excerpt of Selected Focused Concepts.

Code ID	Original Transcript (in Chinese)	Initial Code ID	Initial Coding (in Chinese)	Focused Code ID	Focused Concept
T4-21	“I still hope to receive continuous rehabilitation guidance … and ongoing rehabilitation training.”	T4-a21	Hopes for post-discharge rehabilitation guidance and structured training to improve recovery outcomes.	T4-aa21	Service Content Orientation
T6-26	“More rehabilitation therapists … to correct my movements.”	T6-a26	Expects therapists to provide more individualized guidance to enhance rehabilitation effectiveness.	T6-aa26	Quality of Rehabilitation Service Delivery
T1-29	“The current rehabilitation and nursing services … are just passable.”	T1-a29	Perceives current rehabilitation and nursing services as suboptimal and needing improvement.	T1-aa29	Personalised Collaboration

), and then further abstracted into theoretically meaningful subcategories.

For example, codes such as “recording blood pressure,” “taking medicine on time,” and “regular checkups” were synthesized into focused concepts like “chronic disease management” and “health information recommendations,” which were then categorized under the subcategory “health needs.” Likewise, codes like “difficulty operating” and “need for voice assistance” were grouped under the subcategory “technology adaptation challenges” (see

Table 6. Excerpt of Selected Focused Concepts–Subcategories (Focused Coding).

Code ID	Focused Concept (in Chinese)	Axial Code	Axial Category (Sub-Dimension)
T1-aa2	Health information advice	A1	Health Needs
S1-aa39	Elderly individuals prefer ageing in place	A2	Ageing-in-Place Needs
T8-aa75	Emergency health response	A3	Emergency Support Needs

).

A total of 28 subcategories were generated, covering core aspects such as health management, platform functionality, emotional support, mobility, information acquisition, and cross-module integration. These subcategories served as key intermediaries, leading to the development of major categories and the three-tier model structure.

4.1.3. Axial Coding Phase

Following the establishment of subcategories, the axial coding phase involved organizing and abstracting semantic linkages across subcategories to derive nine logically distinct and semantically stable major categories. The objective was to conceptually consolidate related subcategories into higher-order categories that reflect recurring themes in participant narratives.

Through cross-category semantic comparison and conceptual merging, nine major categories were identified, encompassing the full trajectory from user need expression to platform use expectations and outcome feedback. These categories demonstrated high semantic inclusiveness and logical independence, enabling the unification of heterogeneous expressions across settings and respondents. Compared with subcategories, the major categories increased the level of abstraction and supported horizontal comparison and theoretical model development. Each major category integrated multiple conceptually related subcategories and had both classificatory coherence and practical relevance. The major categories and their corresponding subcategories are summarized in

Table 7. Axial Coding Results: Construction and Hierarchical Integration of Nine Core Categories.

Code	Subcategory (Subdimension)	Code	Core Category (Thematic Domain)
A1	Health-related needs	AA1	Health and Medical Care Needs
A2	Ageing-in-place needs
A3	Emergency support needs
B1	Technology adaptation ability	BB1	Technology Adaptation and Digital Learning
B2	Technology Adaptation Challenges
B3	Technology Adaptation Needs
B4	Technology training needs
C1	Emotional Needs	CC1	Psychological and Emotional Support
C2	Emotional Support
C3	Social Needs
C4	Social Support
C5	Psychological needs
C6	Psychological Support Challenges
D1	Family support	DD1	Family and Social Support
D2	Community support
D3	Challenges in Community Support
D4	Community Support Needs
D5	Community Infrastructure Support
E1	Behavior-related needs	EE1	Daily Life and Mobility Support
E2	Lifestyle habits
E3	Daily living needs
E4	Material support needs
F1	Policy-related barriers	FF1	Policy and Resource Management
F2	Resource management needs
F3	Resource integration needs
G1	Service content orientation	GG1	Digital Platform and Intelligent Service Design
G2	Service design challenges
G3	Service design needs
G4	Platform Design Needs
G5	Resource integration mode
H1	Adaptive Response Mechanism	HH1	Smart Interactive Mechanisms of Digital Platforms
H2	Personalised Coordination Mechanism
H3	Intelligent Matching Mechanism
H4	Needs Expression Mechanism
I1	Social feedback and effectiveness	II1	Effectiveness of Digital Platform Use
I2	Resource response efficiency
I3	Needs satisfaction efficiency

.

4.1.4. Theoretical Coding Phase

1.

Three-Tier Classification Framework

In the theoretical coding stage, the nine major categories were grouped into three overarching levels according to their functional logic and semantic roles: the Needs Layer, Mechanisms Layer, and Outcomes Layer. This classification constituted the final structure of the category system. The three levels form a logical chain that begins with user needs, proceeds through platform mechanisms, and culminates in measurable usage outcomes (see

Table 8. Three-Level Category Names and Their Covered Domains.

Needs Layer	Mechanism Layer	Outcome Layer
Health and Medical Care Needs	Intelligent Interaction Mechanisms of Digital Platforms	Usage Outcomes of Digital Platforms
Technology Adaptation and Digital Learning	(Subdimensions) Adaptive Response Mechanism Personalised Matching Mechanism Intelligent Recommendation Mechanism Needs Expression Mechanism	(Subdimensions) Social Participation Outcomes Resource Allocation Efficiency Needs Fulfillment Efficiency
Psychological and Emotional Support
Family and Social Support
Daily Life and Mobility Support
Policy and Resource Management
Smart Service Design of Digital Platforms

).

2.

Contextual Coding for Layer Verification

To further validate the structural relationships and users’ psychological expectations across the three levels, the study applied “contextual linkage” coding. This focused on common sentence patterns such as “I need … if the platform can …, then I will …,” which capture the direct connections among needs, platform responses, and behavioral intentions. These linguistic structures provided key empirical evidence for the logic of the model. Selected coding examples across the three levels are shown in

Table 9. Excerpt of “Original Text–Needs Level–Mechanism Level–Outcome Level” Coding.

Original Quote (in Chinese)	Corresponding Needs Layer	Corresponding Mechanism Layer	Corresponding Outcome Layer
“When I first used the smart wristband to check my heart rate and step count, it was a bit confusing to set up.”	Technology Adaptation and Digital Learning	Adaptive Response Mechanism	Needs Fulfillment Effectiveness
“Sometimes the scheduling of community health lectures is inconvenient—either overlapping or at the wrong time.”	Health and Medical Care Needs	Adaptive Response Mechanism	Needs Fulfillment Effectiveness
“If the platform could help me book doctor appointments and specialist numbers in advance, it would save me a lot of trouble each time I go to a big hospital.”	Smart Service Design of Digital Platforms	Personalised Matching Mechanism	Resource Optimisation Effectiveness

.

3.

Derivation of Platform Response Logic

These contextual structures not only reflected functional linkages between needs and mechanisms but also illustrated perceived and behavioral outcomes at the result layer. For example:

• “I need tailored advice for my condition→ Health Management Need”
• “The platform should automatically push updates → Smart Matching Mechanism”
• “It makes me feel reassured to know what to do → Perceived Need Fulfillment”

To further theorize these connections, seven contextual logic units were extracted, including “Need Characteristic → Mechanism Selection,” “Platform Mechanism → Perceived Outcome,” and “Dynamic Adaptation → Technological Barrier Mitigation” (see

Table 10. Analysis Results of “Elderly Needs Characteristics and Platform Response Mechanisms.

Analytical Unit	Interpretation of Analytical Content
Need characteristics → Mechanism selection	Based on empirical data, older adults’ health management, technical interaction, emotional support, and behavioral convenience needs may trigger preferences for specific functional mechanisms. This leads to the selection of smart interactive mechanisms. For example, the need for blood pressure monitoring may activate the “simplified processing” function, while “overly complicated interfaces” may reduce usability.
Need types → Mechanism preference	Different types of needs correspond to preferences for different mechanisms. For example, health management may correspond to personalised matching; digital learning needs may activate adaptive response; and social participation needs may align with needs expression and feedback, highlighting the platform’s adaptive allocation capability.
Smart platform mechanism → Need responsiveness	Whether the smart platform’s embedded mechanisms are perceived as “useful” or “helpful” by older adults determines its responsiveness. For instance, a medication reminder feature or device may reduce memory-related health risks. Real-time interaction functions enhance clarity and convenience, thus improving perceived responsiveness.
Adaptive response mechanism → Technical barrier mitigation	For older adults who struggle with digital tools, adaptive response mechanisms can simplify usage through habit-based personalization, streamlined operations, and auxiliary prompts, ultimately reducing interaction burden and increasing motivation.
Intelligent matching mechanism → Accurate service targeting	This mechanism enables the system to analyze health profiles and automatically recommend appropriate services (e.g., dietary suggestions, activity matching), thereby improving life efficiency and increasing trust in the platform’s “lifestyle compatibility.”
Personalised collaboration mechanism → Cross-domain integration support	When needs span multiple domains—e.g., health, transportation, recreation—this mechanism coordinates support from various modules such as booking, escort services, or family co-care, ensuring seamless integration and enhanced service delivery.
Needs expression mechanism → Action-driven interaction	If the platform lacks accessible channels for articulating needs (e.g., one-click assistance, emotional support), older adults may disengage over time. Therefore, facilitating intuitive need expression is key to sustaining user participation.
Smart mechanism use → Perceived effectiveness	If the platform mechanisms function inconsistently, such as redundant services or unclear functions, this undermines perceived efficiency and trust. On the contrary, effective integration improves perceived system reliability and user confidence.

). These served as the foundational logic blocks for the model construction.

4.

Supplementary Analysis of Semantic Co-occurrence

The analysis also identified high degrees of semantic overlap across categories. Some participant statements simultaneously involved multiple need domains (e.g., health and family support), reflecting the integrative and composite nature of real-life contexts. Representative statements and their co-occurring categories were compiled (see

Table 11. Excerpt of “Internal Relationship Analysis Table for Event One Core Category” Coding.

Original Quote (in Chinese)	Relationship Code	Thematic Relationship
“If there were a health platform, I hope it could manage things like medication reminders. My memory isn’t great, and I often forget appointments or which specialist to see at the hospital—it’s really troublesome. Also, if I could consult with a health coach when needed, that would be great for minor issues or mental concerns.”	A digital platform can support health monitoring, reminders, remote consultations, etc., improving healthcare service accessibility and helping older adults manage their health.	Health and Medical Needs ←→ Smart Platform Service Design
“My family really cares for me. They often check in on my physical condition and accompany me to see doctors. This emotional warmth makes me feel less anxious. But sometimes I feel lonely; I wish there were more community activities to talk, share stories, and feel less alone.”	The concern of family and friends has a direct emotional effect; when paired with community support, it helps alleviate loneliness.	Psychological and Emotional Support ←→ Family and Social Support
“Policies should help arrange more convenient transport services for the elderly, like community shuttles or easy access vehicles. Sometimes I don’t even know how to get to the hospital. Policies that consider this can really improve older adults’ ability to move freely.”	Policy design affects older adults’ ability to maintain independent mobility—for instance, through accessible transport or barrier-free environments.	Daily Living and Mobility Support ←→ Policy and Resource Management

) to support the coherence and comprehensiveness of the model’s design logic.

To synthesize these findings and visually represent the category hierarchy and interrelationships, a “Category Hierarchy Structure Diagram” was constructed (see Figure 3), providing a structural overview prior to theoretical model formulation.

4.1.5. Theoretical Saturation Test

Theoretical saturation is achieved when additional data analyzed using the same coding system no longer yield new categories, concepts, or relationships among them [68]. In this study, theoretical sampling and in-depth interviews were conducted simultaneously, with continuous comparative analysis used to refine and expand the category framework.

During the final stage of data collection, grounded coding was conducted on the last three interviewees: a 62-year-old male maintenance technician from Zhuhai, a 66-year-old retired middle school teacher from Foshan, and a 63-year-old female retail employee from Dongguan. No new categories or category relationships were identified in their data, indicating that the theoretical framework developed in this study had reached saturation.

4.2. Constructed Grounded Theory Model: Structure and Explanation

To synthesize the qualitative findings derived from the grounded theory analysis, this section presents a user-centred theoretical model (see Figure 4), developed through CGT. The model delineates a three-tier logical structure—User Needs, Platform Mechanisms, and Service Outcomes—to interpret how older adults’ contextualized needs are translated into concrete service outcomes through interactions with digital platforms. This framework provides theoretical support for subsequent design applications and quantitative validation.

1.

Needs Layer: Core Identification

The first tier, the Needs Layer, encompasses seven core domains of user needs in community-based elderly care settings: ① Health and medical care needs, ② Daily life and mobility support, ③ Psychological and emotional support, ④ Family and social support, ⑤ Technological adaptation and digital learning, ⑥ Policy and resource management, and ⑦ Digital platform and smart service design. This layer represents users’ fundamental service expectations prior to platform interaction and serves as the starting point for the service design process. It emphasizes the necessity for digital platforms to flexibly adapt and structurally configure services according to the diversity of user needs.

2.

Mechanisms Layer: Translating Needs into Responses

The second tier, the Mechanisms Layer, acts as a mediating structure linking user needs to service outcomes. It captures the intelligent interaction strategies deployed by platforms in response to these needs. This layer is conceptualized as the overarching domain of Smart Interactive and Adaptive Mechanisms, composed of four specific sub-mechanisms: ① Need articulation mechanism, ② Dynamic adaptation mechanism, ③ Personalised coordination mechanism, and ④ Intelligent matching mechanism. These mechanisms embody critical platform functions—such as contextual adjustment, information recommendation, and service integration—and are designed to enhance real-time responsiveness, personalization, and sustainability. As such, this layer constitutes the operational core for connecting diverse elderly needs to responsive service delivery.

3.

Outcomes Layer: Fulfillment and Feedback

The third tier, the Outcomes Layer, captures the subjective experiences and behavioral feedback of older adults following their engagement with digital platforms. It comprises three outcome dimensions: ① perceived need fulfillment, ② social participation and identity, and ③ resource optimisation effects. These outcomes reflect whether the platform has successfully translated and fulfilled core user needs while also indicating its potential role in fostering trust, encourageing continued use and strengthening social connectedness. From a service design perspective, this layer highlights the importance of ongoing service refinement through the accumulation of user experience and interactive feedback.

Accordingly, the model also illustrates semantic linkages and interactive relationships across tiers. For example, needs such as” psychological support” and “family support” may influence outcomes not only indirectly via the “Mechanisms Layer” but also directly by triggering “social participation and identity.” Moreover, the category of “digital platform and smart service design” spans both the “Needs and Mechanisms” layers, reinforcing the platform’s foundational role in supporting the underlying service logic. As the core empirical result of this study, the model demonstrates how multilayered elderly needs are classified, addressed, and transformed through platform interfaces. It provides a category framework that is logically structured, contextually grounded, and semantically coherent, offering a solid theoretical foundation for both quantitative validation and design application.

4.3. Model Validation

4.3.1. Questionnaire Design and Data Collection

The questionnaire in this study was developed based on the core and subcategories derived from the CGT analysis, translating abstract qualitative concepts into operationalized quantitative indicators. This served as a critical bridge from theoretical construction to empirical validation [66]. The questionnaire consisted of five major modules: demographic information, multidimensional needs, platform interaction mechanisms, platform use outcomes, and open-ended questions, totaling 110 items.

Each module corresponded to the three-tier structure established through CGT. The first seven modules reflected the seven major categories of the Needs Layer, including: health and medical care, technology adaptation, emotional support, family and social support, daily life and mobility assistance, policy and resource management, and digital platform and smart service design. The final two modules were aligned with the Mechanisms Layer and Outcomes Layer, respectively, and were combined into two integrated modules: “Smart Interaction and Adaptation Mechanisms” and “Platform Use Outcomes.” This integration strategy enhanced both the internal consistency and practical administration of the questionnaire while also facilitating statistical procedures such as exploratory factor analysis and regression modeling without sacrificing the theoretical depth and logical integrity of the original framework [67] (see

Table 12. Correspondence Table Between “Questionnaire Structure” and “Core Categories of the Grounded Theory Model”.

Questionnaire Module	Corresponding Core Category from Theoretical Model	Associated Layer
Health and Medical Care Needs	Health and Medical Care Needs	Needs Layer
Technology Adaptation and Digital Learning	Technology Adaptation and Digital Learning
Psychological and Emotional Support	Psychological and Emotional Support
Family and Social Support	Family and Social Support
Daily Life and Mobility Support	Daily Life and Mobility Support
Policy and Resource Management	Policy and Resource Management
Smart Service Design of Digital Platforms	Smart Service Design of Digital Platforms	Mechanism Layer
Intelligent Interaction Mechanisms of Digital Platforms	Intelligent Interaction Mechanisms of Digital Platforms
Perceived Effectiveness of Digital Platform Use	Perceived Effectiveness of Digital Platform Use	Outcome Layer

).

All items were measured using a five-point Likert scale. The language was carefully designed to be clear and accessible to older adults, avoiding technical jargon and complex grammar structures to improve comprehensibility and data quality. The survey was conducted between February and March 2025 across six cities in the PRD region—Guangzhou, Shenzhen, Dongguan, Foshan, Zhongshan, and Zhuhai—through both online and paper-based distribution. A total of 352 questionnaires were collected, with 322 deemed valid, yielding a valid response rate of 91%.

Among respondents, 51.86% were male and 48.14% were female. The majority (28.88%) were aged 65–69, and most had an education level of junior high school or below. Approximately 27.02% reported frequent use of smart devices, while 47.52% reported using them only occasionally or never, indicating a moderate degree of digital divide. Overall, the sample demonstrated a diverse range of demographic and lifestyle characteristics, making it suitable for subsequent statistical analyses and group comparisons.

4.3.2. Data Analysis Results

1.

Reliability and Validity Testing

To validate the questionnaire as a quantitative instrument for the three-tier model constructed through CGT, this study conducted Cronbach’s α reliability analysis and both KMO and Bartlett’s tests of sphericity for each major construct. The results showed that the Cronbach’s α values across all dimensions ranged from 0.91 to 0.97, indicating excellent overall internal consistency. This suggests a high degree of coherence and stability in measuring the same underlying constructs. Notably, the dimensions of Psychological and Emotional Support Needs and Digital Platform and Smart Service Design Needs reached α values of 0.96 and 0.97, respectively, reflecting particularly strong reliability (see

Table 13. Reliability Test Results of Questionnaire Modules for Each Construct.

Dimension	Cronbach’s Alpha
Health and Medical Care Needs	0.91
Technology Adaptation and Digital Learning Needs	0.95
Psychological and Emotional Support Needs	0.96
Family and Social Support Needs	0.96
Daily Life and Mobility Support Needs	0.95
Policy and Resource Management Needs	0.95
Smart Service Design of Digital Platforms	0.97
Intelligent Interaction Mechanisms of Digital Platforms	0.95
Perceived Effectiveness of Digital Platform Use	0.94

).

For validity testing, all KMO values were at least 0.92, indicating strong sampling adequacy for factor analysis. The module for Digital Platform and Smart Service Design Needs scored the highest at 0.98, denoting an excellent level. Bartlett’s test of sphericity yielded significant results across all constructs (p < 0.001), confirming strong inter-variable correlations and supporting the appropriateness of subsequent structural analyses (see

Table 14. Validity Test Results of Questionnaire Modules for Each Construct.

Dimension	KMO (Overall)	Bartlett’s Chi-Square	df	p-Value
Health and Medical Care Needs	0.92	1240.369	15	3.48 × 10−255
Technology Adaptation and Digital Learning Needs	0.97	2588.731	45	0
Psychological and Emotional Support Needs	0.98	3939.056	91	0
Family and Social Support Needs	0.98	4001.502	91	0
Daily Life and Mobility Support Needs	0.96	2666.347	45	0
Policy and Resource Management Needs	0.96	2521.239	45	0
Smart Service Design of Digital Platforms	0.98	4717.906	120	0
Intelligent Interaction Mechanisms of Digital Platforms	0.97	2518.097	45	0
Perceived Effectiveness of Digital Platform Use	0.95	2022.101	28	0

). These reliability and validity findings demonstrate the questionnaire’s high measurement quality, providing both a statistical basis for exploratory factor analysis and empirical support for the theoretical model.

2.

Exploratory Factor Analysis (EFA) Results

This study employed Exploratory Factor Analysis (EFA) to assess the convergent validity and structural consistency of the questionnaire items under each major construct. The core objective was to examine whether the subcategories derived from CGT could be statistically aggregated into a single latent construct after being operationalized into questionnaire items. If the items load significantly onto a single factor and meet the criteria for factor loadings and communalities, this would provide empirical support for the original qualitative classification logic and lay a foundation for subsequent factor construction and regression validation (see

Table 15. Excerpt of Exploratory Factor Analysis (EFA) Results.

Dimension	Item	Factor Loading	Communality	Meets Threshold
Health and Medical Care Needs	1	0.93	0.89	Yes
Health and Medical Care Needs	2	0.85	0.71	Yes
Health and Medical Care Needs	3	0.85	0.71	Yes
…	…	…	…	Yes

; detailed item-level analysis is presented in Appendix A).

The EFA results revealed that the factor loadings for each construct ranged from 0.79 to 0.88, while communalities mostly fell between 0.64 and 0.76. For the needs layer, the dimension of Health and Medical Needs yielded loadings between 0.83 and 0.86, with communalities around 0.70–0.73; Technology Adaptation and Digital Learning Needs showed loadings of 0.81–0.86 with communalities above 0.65; Psychological and Emotional Support Needs had loadings between 0.80 and 0.85; Family and Social Support Needs ranged from 0.79 to 0.88, both exhibiting communalities around 0.70. The constructs of Daily Life and Mobility Support and Policy and Resource Management Needs yielded factor loadings between 0.81 and 0.86, with communalities between 0.69 and 0.74. Digital Platform and Smart Service Design Needs had loadings mostly above 0.84, peaking at 0.87, with communalities reaching 0.76. As for the mechanism and effect layers, Platform Intelligent Service Mechanism showed factor loadings between 0.81 and 0.85, and communalities between 0.65 and 0.73; Expected Effectiveness of Platform Use yielded loadings between 0.81 and 0.86, with communalities ranging from 0.66 to 0.75. Overall, each construct successfully extracted a single principal factor with no cross-loading, and both factor loadings and communalities met statistical criteria, indicating strong construct validity and structural stability.

3.

Correlation Analysis

Pearson correlation analysis was conducted to examine the relationships between the total scores of the nine major constructs. After excluding missing values, correlation matrices were generated using the mean scores of each construct, followed by significance testing (see Figure 5).

Results showed that all constructs were significantly positively correlated (p < 0.01), with most exhibiting moderate-to-strong linear relationships. This suggests a high degree of interdependence among user needs and platform-related constructs. For example, the correlation coefficient between Technology Adaptation and Digital Learning and Platform Intelligent Service Mechanism reached 0.94, while the correlation between Psychological and Emotional Support and Expected Effectiveness of Platform Use was as high as 0.95, reflecting strong alignment between user needs and expected platform outcomes. In general, the high coupling across constructs empirically supports the multidirectional demand-mechanism linkages identified in the grounded theory model and provides a statistical basis for subsequent regression modeling.

4.

Multiple Linear Regression Analysis

To assess the predictive power of user needs on the usage and perceived benefits of smart elderly care platforms, the study first tested for multicollinearity among the seven needs constructs. Results indicated significant multicollinearity (all VIF values > 5). To address this, Principal Component Analysis (PCA) was employed to reduce dimensionality and consolidate these correlated constructs into a single comprehensive indicator (PCA1), thereby improving model stability and explanatory power. The seven constructs included: Health and Medical Needs, Technology Adaptation and Digital Learning, Psychological and Emotional Support, Family and Social Support, Daily Life and Mobility Support, Policy and Resource Management, and Digital Platform and Smart Service Design Needs. PCA results confirmed that PCA1 captured the majority of variance with balanced factor loadings, making it a suitable integrative indicator of overall user needs among the elderly.

Using PCA1 as the independent variable, two regression models were constructed to predict two core dependent variables: smart service usage (smart_service) and platform use effectiveness (use_effect). Model 1 excluded control variables, while Model 2 included covariates such as age, gender, education level, household composition, frequency of device use, and prior platform experience.

Model 1 results indicated that PCA1 strongly predicted both usage and effectiveness (β = 4.38, 3.53; p < 0.001), with R² values of 0.94 and 0.91, respectively, demonstrating high explanatory power (see

Table 16. Multiple Regression Analysis Results of PCA1 on the Use and Effectiveness of Smart Elderly Care Service Platforms.

Variable	Model 1 β	p-Value	Model 2 β	p-Value
Intercept	32.98	<0.001 ***	32.57	<0.001 ***
PCA1	4.38	<0.001 ***	4.25	<0.001 ***

Note: *** p < 0.001.

).

In Model 2, even after including control variables, PCA1 remained a stable predictor (β = 4.25, 3.39; p < 0.001), with adjusted R² values of 0.94 and 0.92. Among the control variables, only “higher education” and “low frequency of device use” showed marginal or significant negative effects on perceived outcomes, suggesting that technological familiarity and usage stability play key roles in shaping elderly users’ perceptions of platform benefits (see

Table 17. Multiple Regression Analysis Results of PCA1 on Usage and Platform Benefits After Controlling for Background Variables.

Variable	Model 1 β	p-Value	Model 2 β	p-Value
Intercept	26.32	<0.001 ***	27.70	<0.001 ***
PCA1	3.53	<0.001 ***	3.39	<0.001 ***

Note: *** p < 0.001.

).

To summarise, PCA1—as an integrative indicator of the seven needs constructs—effectively addressed multicollinearity issues while providing strong explanatory power for both smart service usage and outcome evaluation. Given the high conceptual coupling between PCA1 and the dependent variables, the resulting high R² values in regression models reflect internal consistency rather than overfitting. These findings offer robust empirical support for the three-tier model of User Needs–Platform Mechanisms–Service Outcomes developed through CGT, confirming its structural coherence and statistical validity in the quantitative phase. Moreover, the results underscore the importance of addressing multidimensional user needs and technology interaction behaviors in smart elderly care platform design and policy advancement in order to enhance service acceptance, user outcomes, and long-term sustainability.

4.4. Integrated Discussion of Qualitative and Quantitative Results

The qualitative and quantitative results of this study jointly support the UN-PM-SO model and further reveal its implications for the design of community-based digital elderly care platforms.

First, the qualitative analysis shows that the interview data clearly converged into a three-tier structure: the needs layer encompasses seven core categories of needs, the mechanisms layer consists of four types of smart interactive mechanisms, and the outcomes layer presents three categories of results—need fulfillment, social participation, and resource optimisation. These findings not only validate the logical chain of “need activation–mechanism mediation–outcome transformation” but also indicate that older adults’ digital usage behaviors are not isolated events; rather, they are driven by needs, translated through mechanisms, and ultimately generate perceivable benefits. Second, the quantitative examination further confirmed the stability of the model. Reliability and validity tests reached satisfactory levels, and EFA demonstrated clear factor aggregation consistent with the theoretical framework. Given the high multicollinearity among the seven needs dimensions, this study employed PCA1 as an indicator of “overall demand intensity.” Regression analysis revealed that overall demand intensity exerted a significant positive effect on both smart elderly care service usage and evaluations of platform effectiveness. This relationship remained robust even after controlling for demographic characteristics and prior usage experiences. These findings highlight the dominant role of needs in shaping platform usage and confirm that effective mechanism design can substantially enhance perceived effectiveness.

Taken together, the qualitative and quantitative results form complementary validation: the needs layer reveals the drivers of use, the mechanisms layer establishes the pathways of conversion, and the outcomes layer provides measurable feedback indicators. More importantly, the study shows that if needs are not accurately captured or lack corresponding mechanisms, even with sufficient access conditions and high-frequency usage, the perceived benefits remain fragile and the chain is disrupted. Conversely, when needs are timely translated and effectively linked through mechanisms, sustained benefits can emerge even from low-frequency participation.

This finding not only validates the chain logic of the UN-PM-SO model but also offers actionable pathways for platform design—need alignment, mechanism mediation, and outcome transformation—indicating that design should prioritize contextual adaptation and the management of behavioral frictions rather than simply expanding functional modules. Therefore, the integrated results presented above not only verify the theoretical chain of the UN-PM-SO model but also suggest practical directions for platform design.

5. Discussion

To further elaborate the theoretical value and practical potential of the User Needs–Platform Mechanisms–Service Outcomes three-tiered model proposed in this study, this section discusses its contributions from three perspectives: (1) Theoretical positioning—by engageing in systematic dialogue with existing theories such as technology acceptance, human–computer interaction (HCI), recommender systems, and service experience, this section illustrates how the proposed model complements and extends existing frameworks while clarifying its theoretical position and innovative implications within the field of smart ageing; (2) Translation into platform design strategies—based on the behavioral logic and structural mechanisms revealed by the model, a series of actionable digital platform design strategies and modular proposals are presented, serving as a bridge between theoretical construction and practical application; and (3) Policy and governance implications—this section explores the model’s potential for integration in fragmented elderly care governance scenarios and its implications for multi-sectoral collaboration, need-oriented policy formulation, and digital service governance systems. The subsequent sections expand on each of these perspectives to demonstrate the model’s integrative capacity and practical relevance in the context of ageing services, addressing the systemic challenges long embedded in the design of digital ageing platforms.

5.1. Theoretical Positioning and Literature Dialogue of the Grounded Model

This study proposes the User Needs–Platform Mechanisms–Service Outcomes model based on CGT to address the lack of integration among user context, structural conditions, and design logic in the field of smart ageing digital services. Through interviews and contextual analysis, the findings reveal that older adults’ platform usage behavior is influenced not only by their attitudes toward technology but also by an interplay of factors such as health status, resource accessibility, social support, and policy incentives, highlighting the limitations of conventional technology acceptance theories in ageing contexts.

At the user needs layer, this study identifies seven core domains: health, psychological well-being, family support, mobility, technology, and resource access. While these partially align with constructs such as “perceived usefulness” and “perceived ease of use” in the TAM model [69], the findings suggest that older adults’ technology acceptance is more context-driven and strategic, rather than based on isolated functionality. Variables such as “social influence” and “facilitating conditions” in the UTAUT model [70] offer some explanatory power but fail to encompass the multifaceted decision-making conditions uncovered here, including family assistance, community support, alternatives, and policy subsidies [71,72]. The proposed model further emphasizes the contextual dependency and multi-source generation logic of user needs, addressing the contextual blind spots of mainstream technology acceptance research.

At the platform mechanism layer, four key mechanisms are proposed: need articulation, dynamic adaptation, intelligent matching, and personalised coordination. These align with HCI principles such as perceived control and real-time feedback [73] and echo the recommendations by Sonboli on the importance of explainability and fairness in recommender systems for older adults [74]. To mitigate cognitive overload caused by excessive information, platforms should strike a balance between recommendation transparency and mental effort [75]. Additionally, respondents expressed strong expectations for collaborative coordination, suggesting that recommender systems should evolve toward hybrid interaction models that integrate data reasoning with human assistance—supporting multi-agency cooperation and the building of trust among elderly users.

At the service outcomes layer, the model reveals three major psychological effects triggered by platform engagement: perceived fulfillment, resource optimisation, and social participation and identity. These experiences stem not only from service outcomes but also from responsiveness, process simplicity, and the feeling of being understood, corresponding to SERVQUAL indicators of responsiveness, empathy, and reliability [76]. Social interaction and self-identity also emerge as crucial benefits. Many participants indicated that platforms facilitating community involvement and emotional connection significantly strengthen sustained usage intention. These findings align with studies on the role of social networks in ageing mental health and the positive impact of connectedness on well-being [77].

To summarise, the three-tiered model constructed through grounded theory not only addresses the limitations of TAM, UTAUT, and recommender systems in ageing contexts but also offers a semantically structured perspective linking need generation, platform response, and subjective psychological effects. This model serves as a theoretically robust framework for digital platform design while laying a structural foundation for interdisciplinary integration in the smart ageing field, embodying a theory of need-orientation, contextual sensitivity, and systemic integration. At the same time, the value of this model does not lie in requiring comprehensive implementation all at once but rather in providing a decomposable logical framework that allows practices in different contexts to be advanced step by step according to available resources and governance needs. In other words, its design intent is closer to a modular and phased guiding structure, whereby even the application of only partial modules can yield tangible demand responses and improvements in outcomes. This characteristic also lays the foundation for the subsequent translation into design strategies.

5.2. Model-Driven Transformation into Design Strategies

Building upon the Needs–Mechanisms–Effects three-tiered theoretical model developed in this study, a set of digital platform design strategies was derived to translate abstract behavioral structures from the qualitative analysis into actionable and evaluable service modules (see Figure 5). This strategy framework not only reflects a structured understanding of elderly user needs but also demonstrates how the model supports cross-scenario integration, intelligent interaction design, and optimisation of user experience, highlighting the theoretical-to-practical translational value of CGT in digital health platform development.

At the needs layer, the model suggests that digital platforms should be oriented around the core needs of elderly users and built upon modular and adaptable service structures. As such, platform design must encompass the seven major need dimensions identified in this study, including health management, emotional support, social interaction, home care, and resource linkage. Through smart dispatching and dynamic matching technologies, the platform should deliver personalised and context-sensitive service responses. This strategy addresses concerns in existing research regarding the “integration challenge of heterogeneous services,” advocating for platforms to shift away from function-stacking architectures toward integrated coordination across modular components, thus offering elderly users a smoother and more sustainable usage experience.

In the mechanisms layer, the corresponding design strategies emphasize the platform’s ability to coordinate across modules and perceive context in order to support the complex logic of user needs observed during interaction. Results show that elderly users are highly sensitive to adaptability in interfaces and response mechanisms. Hence, the platform must enhance interaction friendliness and system responsiveness, with specific strategies including the implementation of voice assistance, visual guidance, and simplified operations to reduce learning thresholds and improve cognitive accessibility. Additionally, the platform should feature smart adaptation systems capable of adjusting automatically to health status, lifestyle rhythms, and usage behaviors. This study also emphasizes that the transparency and explainability of platform behavior significantly influence continued use. Accordingly, personalised recommendations and intelligent matching modules should be integrated to enhance trust and predictive capacity in elderly-platform interaction.

At the effects layer, the study reveals that elderly users perceive the platform’s value not only in terms of functional satisfaction but also in extended benefits related to subjective experience and social participation. Therefore, platform design must emphasize the construction of perceived value, especially by incorporating modules that support social engagement and emotional companionship to enhance users’ psychological well-being and social connection. Furthermore, remote psychological counseling and crisis intervention services should be embedded to respond to expressed needs around loneliness and safety. To ensure the platform’s long-term sustainability, a dynamic feedback and service evaluation system should be established to incorporate user satisfaction and suggestions into iterative improvements, thus enabling a “use–adjust–reuse” feedback loop for continuous system optimisation.

In addition, this study emphasizes that design strategies should not be implemented in full all at once but advanced progressively according to the urgency of service pain points, technological feasibility, and governance costs. Prior studies note that in large and complex health systems, digital tools often fail to achieve long-term sustainability and may even generate high costs and low efficiency without clear prioritization and phased planning [78]. Accordingly, three tiers of implementation are proposed: (1) the basic version, prioritizing need articulation and intelligent matching to resolve the urgent problem of “inaccessible services and delayed responses” with relatively low thresholds [79,80]; (2) the extended version, adding dynamic adaptation to strengthen service continuity and risk prevention; and (3) the integrated version, incorporating personalised coordination to enable cross-organisational resource integration and governance upgrading. These stages correspond to sequential improvements in accessibility, continuity, and governance, offering a progressive configuration from low threshold to high complexity.

Notably, even partial implementation can generate substantial benefits. Initial deployment of need articulation and intelligent matching improves service accessibility and timely responses. With dynamic adaptation, platforms adjust to changing health conditions or daily routines, enhancing stability and safety. The addition of personalised coordination achieves cross-sector integration, yielding higher-level optimisation and governance benefits. Stepwise implementation, therefore, ensures cumulative improvements—progressing from basic accessibility → to service stability → to overall governance—while producing observable outcomes at each stage and building momentum for stakeholder collaboration (see Section 5.3). This modular, phased approach also aligns with international guidelines on digital health platforms, which emphasize reusable modules and staged deployment to reduce costs and gradually release system-level benefits [81].

In summary, the strategy framework proposed in this study embodies a comprehensive translation of the model’s theoretical logic—from need emergence to mechanism operation and outcome generation. It directly addresses the complexity and dynamism encountered in real-world applications. Platform design must simultaneously respond to the changing needs of older adults, ongoing technological advancement, and the contextual specificity of service scenarios. Based on this, the study proposes a technical framework grounded in modularity, scalability, and intelligence, with recommendations to gradually integrate smart health devices, telemedicine interfaces, and intelligent home modules to enhance system resilience and self-optimisation capacity. This design logic aligns with international smart care initiatives such as the AGE-Care platform, which similarly emphasizes that the key to elderly-friendly digital platforms lies not in individual technological innovations but in the systematic integration of design strategies driven by user need logic [82]. By translating the model into concrete design strategies, this study deepens the understanding of digital service mechanisms for older adults and provides a theory-informed, practice-oriented design framework for future platform development, addressing prior gaps in strategic derivation and structural alignment and offering both theoretical contribution and applied potential.

5.3. Application from a Stakeholder-Collaborative Perspective

The concept of “stakeholders” originated in the field of business management and was systematized by R. Edward Freeman in the 1980s through his seminal work Strategic Management: A Stakeholder Approach, sparking extensive academic discourse on stakeholder theory. Stakeholders are generally defined as individuals or groups who can affect or are affected by the achievement of organisational objectives [83]. The Needs–Mechanisms–Effects model constructed in this study presents a digitally mediated governance structure, from which a potential multi-actor collaboration mechanism and chain of responsibilities can be inferred. This offers a theoretical foundation for the future implementation and governance design of digital ageing platforms.

However, stakeholder composition and interaction logic are not universally fixed but must be redefined based on regional social and institutional contexts. In the PRD region of China—characterized by high urbanization rates, intensive population mobility, and a market-oriented governance model—digital ageing platforms exhibit a region-specific, multi-level stakeholder structure (see Figure 6). The diversity of elderly populations leads to highly heterogeneous needs, requiring support from frontline participants (e.g., family caregivers, community staff) to assist with platform operation and input of needs. Concurrently, due to the region’s high concentration of technological and commercial resources, backend support actors (e.g., platform enterprises, medical institutions, logistics providers) are responsible for module execution and data processing. As a national pilot area for smart ageing policies, external governance actors (e.g., local governments and subdistrict offices) also participate in platform governance through data integration and policy facilitation. This tripartite structure is not merely a theoretical construct but reflects the interwoven policy drivers, industrial involvement, and demographic shifts specific to the PRD, highlighting the region’s unique governance complexity and demand for systemic integration in ageing practices.

In this densely developed smart community region, with a highly diverse elderly demographic, the digital platform serves as a central mediating node, integrating heterogeneous data sources, coordinating service resources, and generating policy insights. At the needs layer, inputs such as health data, preference settings, and user feedback from elderly individuals initiate service operations. Given disparities in digital literacy and contextual expression, the platform relies on family members to supplement background data and assist with operations, while community workers and property staff contribute observed contextual details that may be difficult for older users to articulate, thus forming the context input interface. At the mechanisms layer, the platform integrates this input and activates corresponding modules (e.g., care dispatch, supply delivery, platform notifications), coordinating caregivers, third-party providers, and logistics networks. Due to the region’s marketized environment, platforms must facilitate high-frequency interactions among enterprises, property managers, care providers, and distributors, forming a dynamic, tightly coupled service network. Continuous information flow with the community ensures service continuity and feedback loops. At the effects layer, task and behavioral data are synthesized to generate user segmentation profiles, module optimisation recommendations, and site-based analytics, which serve both platform adjustment and policy formulation. As a national pilot zone, local governments in the PRD increasingly rely on platform-generated data to allocate resources and design regulations, realizing a closed-loop governance system from user needs to policy response.

As shown in Figure 7 and Figure 8, the digital platform is centred on older adults and forms a dynamic network through the multi-level participation of families, communities, service providers, and government actors. On this basis, both qualitative and quantitative results indicate that stakeholder engagement is sustained not merely by institutional requirements but by tangible benefits generated through platform operation. Specifically, older adults serve both as demand articulators and beneficiaries, with the platform enhancing the timeliness and accessibility of services while strengthening safety and participation through dynamic monitoring and social modules. Caregivers benefit from reduced burdens as caregiving tasks become standardized and traceable. Community workers and property managers improve efficiency and transparency through standardized interfaces. Medical institutions and third-party enterprises leverage stable demand and accumulated data to promote precision services and business innovation. Logistics and service providers increase delivery efficiency and reduce resource waste through real-time dispatching. Government and external governance actors use platform data to optimise policy and resource allocation. Thus, modular and phased implementation ensures both technical and institutional feasibility while providing sustained incentives for multi-stakeholder participation, forming the basis for collaborative governance.

Importantly, such collaboration does not require all actors to be present from the outset but unfolds progressively alongside the phased implementation of modular design. In the core module stage (need articulation and intelligent matching), older adults, family members, and community workers form a basic loop ensuring service access. In the expansion stage (dynamic adaptation), the involvement of medical and logistics actors strengthens continuity and professionalism. In the integration stage (personalised coordination), deeper participation from government and platform enterprises enables cross-sector collaboration and data governance, establishing systemic support. In other words, the platform’s digital design logic—progressing from core to extended to integrated modules—corresponds to the stakeholder participation logic: from accessibility on the demand side, to specialization on the service side, and, ultimately, to institutionalization and sustainability on the governance side. This bidirectional mapping of design and collaboration not only demonstrates the feasibility of phased platform advancement but also highlights the cumulative benefits of dynamic stakeholder engagement.

Therefore, by mapping actual resource flows and service delivery patterns within platforms in the PRD, the above stakeholder architecture can be conceptualized as a multi-dimensional flow system: information flow, service flow, and material flow are coordinated and synchronized through the platform. All actors operate around common data nodes, enabling task triggers and accountability handovers. For example, when an elderly user requests both material delivery and companionship support, the platform can simultaneously notify logistics providers, community care workers, and family caregivers to jointly arrange the response. The user then evaluates service satisfaction, and this feedback feeds into the next cycle of task allocation and module refinement. This platform-mediated logic shifts the governance process from department-centred to demand-driven and shapes a new model of responsive and data-guided care governance.

5.4. Multidimensional Sustainability Value of the Theoretical Model

The UN-PM-SO model constructed in this study reveals the operational logic of community-based digital elderly care platforms and highlights their potential sustainability value, though the related benefits still require gradual validation in practice.

This stepwise design inherently embeds the logic of sustainability: modularisation ensures scalability, dynamic adaptation reduces resource waste, and personalised coordination strengthens governance resilience (see Figure 9). First, at the social level, the needs layer (health, learning, emotional, and social support) and the outcomes layer (demand fulfillment and social participation) demonstrate the platform’s potential to narrow the digital divide and enhance equitable access to services. The prioritized implementation of need articulation and intelligent matching provides a concrete pathway for timely responses to older adults, theoretically aligning with SDG 3 (Good Health and Well-being) and SDG 10 (Reduced Inequalities). Second, at the economic and governance level, the mechanisms layer (“intelligent matching-dynamic adaptation-personalised coordination”) illustrates that a modular and phased deployment can reduce governance and operational costs while improving resource allocation efficiency and cross-sector collaboration across medical, community, and logistics services. This logic is consistent with the proposed design strategies and aligns with SDG 11 (Sustainable Cities and Communities). Third, at the environmental level, although this study does not directly quantify the impacts, remote health services and home-based service scheduling are expected to improve efficiency of access while reducing redundant procedures and resource use, echoing the proposed strategies of “remote health services” and “home service dispatching.” These potential environmental benefits should be considered as directions for future empirical research rather than immediately verifiable outcomes.

In addition, the modular and phased advancement strategies, together with multi-stakeholder collaboration mechanisms, provide an institutional basis for iterative development and platform evolution. However, practical implementation may still face challenges such as technological immaturity, resource coordination difficulties, and governance costs. Thus, the sustainability value of the model should be understood more as a developmental direction and policy goal than as an outcome that can be realized instantly.

6. Conclusions

This study takes the PRD region in China as an empirical setting, focusing on the design challenges and demand logic of digital community-based elderly care platforms. A three-tier theoretical model—comprising Needs, Mechanisms, and Effects—was constructed through a CGT approach. Nine major categories were extracted from interview data and systematically integrated into a three-level classification system with clear contextual translation and behavioral prediction logic. This model was then transformed into a structured questionnaire and empirically validated using Exploratory Factor Analysis (EFA) and multiple regression analysis. Based on these results, a series of practical digital platform design strategies was proposed. Overall, the study emphasizes a demand-driven logic that bridges technical design, contextual adaptability, and service integration, aiming to establish an intelligent elderly care framework characterized by theoretical coherence, practical relevance, and governance potential.

• Theoretical and Content Innovation

The primary innovation of this study lies in its user-centred perspective that integrates dual logics from service design and digital governance. It is among the first to propose a construction framework for digital elderly care platforms, starting from the actual needs of older adults. In terms of design thinking, this research moves beyond the conventional function-stacking approach prevalent in existing platforms and instead develops a modular design logic driven by a multi-level needs structure. It emphasizes that platform mechanisms should dynamically respond to the complex needs of older users, allowing for adaptive customization and contextual scalability.

Additionally, the study introduces a contextual linkage analysis method grounded in CGT, innovatively analyzing recurring narrative structures—such as “I need… if the platform can… then I will…”—to extract and classify theoretical categories. This demonstrates the potential to derive structured models from narrative-based contexts. Furthermore, the platform is positioned as a digital governance intermediary, incorporating stakeholder engagement and data flow mechanisms to systematically illustrate a multi-level governance architecture encompassing information flow, service flow, and policy feedback. This provides an actionable schema for defining the governance function of digital platforms in community-based elderly care systems.

• Theoretical Contributions

On the theoretical front, this study makes several key contributions. First, the Needs–Mechanisms–Effects model built through CGT distinguishes itself from static classification or function-driven frameworks by capturing the dynamic generation of needs and context sensitivity, better reflecting the real-life experiences and decision-making processes of older users during service interactions. Second, the study offers a critical supplement to mainstream technology acceptance models (e.g., TAM, UTAUT), highlighting their limited explanatory power for older populations due to insufficient consideration of family support, policy incentives, and alternative resource contexts. By contrast, the proposed three-stage structure incorporates these contextual factors, enhancing the theoretical scope and interpretability of acceptance models.

Moreover, the research bridges user journey mapping, touchpoint analysis, and intelligent interaction strategies from service design with recommendation logic from digital health platforms, thereby offering a practice-oriented, structurally complete, and cross-context applicable theory–practice framework. This supports a transition from fragmented function-based design to systematic, needs-driven architecture in the field of smart elderly care.

• Practical Contributions

The proposed “Needs–Mechanisms–Outcomes” model not only carries theoretical significance but also demonstrates clear practical potential. The model can be translated into a phased design and implementation pathway: in the basic stage, priority can be given to implementing the “need articulation” and “intelligent matching” modules to directly address older adults’ core pain points of “inaccessible services and delayed responses”; in the extended stage, the addition of the “dynamic adaptation” mechanism can enhance the continuity and safety of the platform across diverse service scenarios; and in the integrated stage, the inclusion of “personalised coordination” enables cross-sector governance and long-term sustainable operation. Importantly, even partial implementation of the modules can significantly improve service accessibility and efficiency, reduce the burden on caregivers, and foster collaborative benefits among communities and enterprises. Therefore, the findings of this study can serve directly as an operational reference for the design of digital elderly care platforms and the formulation of relevant policies.

• Sustainability Value

The theoretical model proposed in this study also embeds the logic of sustainable development. Through modular design and phased deployment, the platform can avoid one-time high costs and low efficiency; through dynamic adaptation and personalised coordination, it can maintain long-term resilience and self-adjustment capacity; and through the participation and collaboration of multiple stakeholders, it can lay the foundation for institutionalized governance. At the social level, the platform helps to narrow the digital divide and enhance service equity (corresponding to SDG 10); at the health level, it promotes the well-being and health management of older populations (corresponding to SDG 3); and at the urban and governance level, it improves resource allocation efficiency and cross-sectoral collaboration (corresponding to SDG 11). Therefore, this study not only offers theoretical contributions and design strategies but also provides a concrete and feasible pathway for the sustainable development of smart elderly care platforms.

• Limitations

Despite the construction of the Needs–Mechanisms–Effects model and corresponding design strategies, this study has several limitations.

First, the study sample was drawn from six cities in China’s Pearl River Delta (PRD). Although the overall scale was relatively limited, it encompassed multiple cities and diverse user groups. For the purposes of CGT and the quantitative validation conducted in this study, the sample size was adequate to ensure theoretical saturation and meet statistical requirements, thereby supporting the model’s construction and testing. As the PRD is a typical region marked by both rapid ageing and digitalization, the findings hold strong representativeness within this context. Nevertheless, their generalization to rural areas or other regions requires further empirical verification. Second, most respondents were self-sufficient and had basic digital literacy, leaving underrepresented groups such as those with disabilities, low education, or low income. These exclusions may reflect researcher subjectivity or potential omissions in category identification.

Third, platform design strategies remain at the theoretical stage without prototype construction or field testing; their usability and interface adaptability—especially for low-tech user groups—remain unverified. Fourth, the study’s quantitative validation relies primarily on EFA and regression analysis, which are limited in capturing nonlinear relationships, mediating effects, or moderation interactions. Self-reported data may also introduce subjective bias or comprehension variation. Lastly, the study does not yet address long-term platform usage or track service outcomes, preventing an assessment of the model’s robustness and adaptability under changing needs or evolving contexts.

• Future Research Directions

To expand the applicability and practical value of the proposed model, future research can proceed along five key directions:

• Sample expansion and cross-context validation: Future research should extend beyond Pearl River Delta cities to central, western, and rural areas and compare elderly care settings such as home-based care, community day-care centers, and integrated medical-care institutions. Samples should also include older adults of different age cohorts, family caregivers, and professional service providers to assess the model’s applicability, stability, and generalizability across diverse socio-cultural and service contexts.
• Prototype and Usability Testing: Translate the theoretical model into platform prototypes and conduct user-centred design testing. Although prior works such as Cristiano have outlined initial functional structures for smart health platforms, they lack integrated multi-level needs logic, which this study seeks to supplement [84].
• Behavioral Data and Longitudinal Tracking: Incorporate behavioral data and longitudinal methods to assess how platform mechanisms affect user experience and needs fulfillment. Sumner has identified the potential for smart platforms to enhance elder participation and self-management, though empirical evidence remains limited [85].
• Cross-Module and Service Integration: Strengthen platform responsiveness by developing strategies for inter-module coordination and scenario-switching, particularly across health management, social interaction, and care services, and explore the feasibility of multi-stakeholder governance collaboration to promote joint participation from government, communities, and enterprises.
• Inclusive Design for Bridging the Digital Divide: Address digital inequalities by improving interface design, assistive technologies, and educational support, creating a more inclusive and user-friendly environment that increases elderly participation and confidence.

Taken together, future research should focus on the long-term operation and sustainability value of the platform. By conducting field testing and behavioral tracking, it will be possible to evaluate its stability and adaptability under evolving needs and changing contexts, thereby advancing smart elderly care platforms from theoretical conception toward efficient and sustainable real-world practice.