Structure–Behavior Coordination of Age-Friendly Community Facilities: A Social Network Analysis Model of Guangzhou’s Cases

Abstract

Rapid population aging calls for a shift from static facility configuration toward understanding how spatial structures coordinate with everyday behavior. This study develops a structure–behavior coordination framework to examine how the spatial embedding of community service centers and surrounding facilities aligns with older adults’ mobility and activity chains. Using Guangzhou as a case, three representative facility aggregation forms—clustered, linear, and patchy—were identified through POI-based spatial analysis. Behavioral mapping supported by Public Participation GIS (PPGIS) and social network analysis captured facility co-use and path continuity, while rank-based measures (Rank-QAP and Rank-Biased Overlap) evaluated correspondence between structural and behavioral centralities. Findings show form-sensitive rather than typological coordination: the clustered case (FY) exhibits compact, mixed-use integration; the linear case (DJ) requires ground-level access along main pedestrian corridors; and the patchy case (LG) relies on a few highly accessible dual-core nodes where improved connectivity strengthens cohesion. Everyday facilities such as markets, parks, and plazas act as behavioral anchors linking routine routes. The framework offers a transferable tool and comparable metrics for diagnosing alignment between built structure and everyday behavior, guiding adaptive, evidence-based planning for age-friendly community systems.

1. Introduction

Rapid population aging and evolving governance are reshaping China’s community public services. Community service centers are widely promoted as walkable platforms that integrate functions and organize neighborhood spaces, while surrounding daily-life facilities provide destinations that support mobility and social participation. Yet whether the spatial structure of facilities aligns with older adults’ observed activity-chain use across different aggregation forms remains unclear. This study examines structure–behavior coordination at the facility scale and introduces a diagnostic framework linking a facility network to a behavioral network, drawing on Guangzhou cases within an aging-in-place policy context.

1.1. Community Service Centers as Platforms for Spatially Embedded Governance

The idea of the community center as an organizing node within a walkable catchment trace to Perry’s neighborhood unit, which clustered schools, parks, and shops to support local identity and everyday interaction (Perry, 2015) [1]. In the United States, postwar neighborhood planning evolved from design-oriented concepts into civic infrastructures and governance frameworks that foster social interaction and local service coordination (Rohe, 2009) [2]. In Singapore, the neighborhood center has evolved into a multifunctional civic complex, integrating healthcare, eldercare, childcare, retail, and recreational spaces within a single, accessible node (Hee & Heng, 2003) [3]. These centers exemplify a decentralized model of urban governance that balances state coordination with community participation, demonstrating how spatially integrated hubs can enhance efficiency and inclusiveness.

In China, community service centers followed a distinct trajectory shaped by communities’ dual roles as administrative units and service providers (Xu et al., 2005) [4]. After housing marketization and welfare decentralization in the 1990s, facility layouts became fragmented and neighborhood-based services weakened (Yang & Zhao, 2002) [5]. To address these challenges, the government initiated a new generation of community service hubs, including the Community Comprehensive Service Centers and Embedded Community Complexes (GB 50180-2018, 2019; The General Office of the State Council of the People’s Republic of China, 2023) [6,7]. These policy frameworks aim to consolidate social resources into “one-stop platforms” that integrate eldercare, childcare, meal provision, and other essential functions (Wang et al., 2023) [8].

Although policy initiatives aim to make service centers anchor points for resource integration, their spatial embedding often varies with urban form, land constraints, and population profiles. Some are poorly connected to surrounding facilities, lack continuous behavioral pathways, and have limited institutional activation. Most studies emphasize institutional frameworks or facility standards, with little focus on their role as embedded service nodes and behavioral hubs. In aging societies, an integrated spatial–behavioral perspective is needed to assess how spatial embedding within different aggregation structures affects their capacity to support older adults’ mobility and well-being.

1.2. Older Adults’ Mobility and Spatial Configuration of Community Facilities

Global age-friendly agendas, particularly the WHO Age-friendly Cities framework, emphasize aging in place, proximate multipurpose destinations, walkability, and opportunities for social participation, articulated across eight interconnected domains that link the built environment with participation and access and that call for neighborhood-scale improvements enabling everyday mobility and interaction (World Health Organization [WHO], 2007, 2015, 2021) [9,10,11].

Empirical studies show that older adults’ daily travel is high-frequency and short-distance (often within 500–800 m), and that accessibility and connectivity in this range shape convenience, activity levels, and social participation (Huang et al., 2019; Wang & Lee, 2010; Keskinen et al., 2019; Grant et al., 2010) [12,13,14,15]. Common multi-purpose destinations—parks, small retail, eateries, and shopping centers—support repeated use and neighborhood interaction (King et al., 2003; Chudyk et al., 2015; Michael et al., 2006; Franke et al., 2017) [16,17,18,19]. Facility diversity and amenity mix are associated with mobility and well-being benefits (Glass & Balfour, 2003; Rosso et al., 2013; Forsyth & Molinsky, 2021; Jayantha et al., 2018) [20,21,22,23].

Beyond coverage and service radii (Bai et al., 2024) [24], facility combinations and aggregation forms condition short-range routes and co-use chains. Clustered or patchy layouts influence route continuity, stop selection, and interaction modes (Wang et al., 2022; Fan et al., 2024) [25,26], while comparative activity-space evidence finds built-environment characteristics exert stronger effects on daily activity patterns than many socioeconomic factors (Zheng et al., 2022) [27]. Density can reduce travel costs and support interaction (Carruthers & Ulfarsson, 2003; Talen, 2006) [28,29], but over-clustering risks redundancy and crowding (Zolfani et al., 2023) [30]. Systematic, quantitative evaluation of structure–behavior coordination at the facility level remains limited, motivating the diagnostic approach adopted here.

1.3. Methodological Approaches to Structure–Behavior Analysis

We denote structure–behavior coordination as the correspondence between a facility’s structural prominence in a facility network and its salience in observed activity chains of older adults. This construct links built form and everyday practice by asking whether spatial configurations that concentrate opportunities are also reflected in the sequences of destinations that people actually use, extending person-centric activity-space and life-circle thinking (Schönfelder & Axhausen, 2003) [31], grounding it in behavioral geography’s focus on environment–mobility relations (Golledge & Stimson, 1997) [32].

At the community scale, empirical approaches offer complementary lenses. Self-reported instruments (diaries, questionnaires) provide contextual detail but face recall bias and limited temporal resolution (Axhausen et al., 2002; Stopher & Greaves, 2007) [33,34]. Passive tracking (GPS/LBS/phone signaling) captures longer movements yet is less reliable for fine-grain, facility-level use due to indoor signal loss and mode ambiguity (Shoval & Isaacson, 2006; Cottrill et al., 2013; Ahas et al., 2010) [35,36,37]. PPGIS-assisted behavioral mapping integrates participant mapping and on-site recording to document activity locations and sequences at facility resolution (Brown & Kyttä, 2014; Kahila-Tani et al., 2019; Shilon & Eizenberg, 2024; Mehrotra et al., 2025) [38,39,40,41]. Building on trip-chaining ideas, activity chains link sequential facility uses into multi-node paths that reveal high-frequency nodes, functional combinations, and continuity effects (Primerano et al., 2008) [42].

These chains form network datasets, with facilities as vertices and sequential usage relationships as weighted edges, enabling structural analysis through social network analysis (SNA). SNA is widely applied in social interaction and health behavior research (Jeong et al., 2024; Ko & Song, 2025) [43,44], and studies demonstrate its ability to translate spatial configurations into interaction patterns, identify key nodes, and guide optimization (Yuan & Zhou, 2025) [45]. However, this potential has not been fully realized, as its application to facility-behavior networks in community public service contexts remains limited.

1.4. Research Need and Objectives

Existing studies have paid limited attention to how the spatial embedding of community service centers within varied facility aggregation structures interacts with older adults’ actual mobility patterns. Prior work has treated facility provision and behavioral analysis separately, leaving the spatial–behavioral correspondence under-specified. We therefore adopt an empirical, diagnostic framework that couples fine-scale behavioral evidence with structural analysis to assess structure–behavior coordination in age-friendly community systems. Using Guangzhou cases, we combine PPGIS-assisted behavioral mapping with paired facility and behavioral networks to assess cross-network alignment across aggregation forms.

The study addresses three main research questions:

RQ1: What facility aggregation forms characterize different types of communities, and how do community service centers embed within these contexts to influence older adults’ mobility and usage patterns?

RQ2: What are the characteristics of high-centrality nodes within behavioral networks, and which traits enable service centers to become high-frequency behavioral hubs?

RQ3: How can empirical evidence on alignment and node roles be translated into resource-aware prioritization guidelines for strengthening coordination between facility networks and observed use?

2. Study Design

2.1. Study Area

Guangzhou, located in southern China, has a permanent population of nearly 19 million (Guangzhou Statistics Bureau, 2025) [46] and encompasses a wide spectrum of community types—from historical neighborhoods and mature residential districts to newly developed zones. As one of the earliest cities in China to pilot community eldercare facilities in the mid-1990s, Guangzhou’s system evolved from the “Starlight Homes for the Elderly,” focused mainly on recreation, to government-funded and NGO-operated Family Comprehensive Service Centers integrating administrative, care, dining, and social functions (Zhu et al., 2018) [47]. By 2022, Guangzhou had established 2759 community service centers, achieving full neighborhood coverage (Guangzhou Civil Affairs Bureau, 2022) [48]. This mature service network, coupled with significant variation in facility aggregation patterns and neighborhood aging levels, provides an ideal context for examining spatial–behavioral coordination between service centers and older adults’ activity networks.

2.2. Research Framework

The framework (Figure 1) comprises four steps.

2.2.1. Community Facility Aggregation Pattern Identification

Point-of-interest (POI) data for 2023 were obtained from Amap, covering nine categories of daily-life facilities (e.g., dining, shopping, healthcare, education, community services, parks), totaling ~420,000 records. Facility clustering was mapped via kernel density estimation, and multiple spatial indicators were integrated into a composite surface using the CRITIC weighting method (Diakoulaki et al., 1995; Okabe et al., 2009) [49,50]. CRITIC quantifies both contrast intensity and inter-correlation across indicators, up-weighting informative dimensions and down-weighting redundancy. Compared with entropy weighting (Liu et al., 2024) [51], it incorporates cross-indicator correlation, giving a more complete picture of multidimensional variation; unlike AHP (Aksu & Küçük, 2020) [52], it avoids subjective scoring, which suits large-scale POI-based analysis.

Based on the CRITIC-weighted surfaces, facilities were classified into aggregation patterns. Case communities were then selected by three criteria: (1) a clear spatial signature representative of one pattern; (2) a full-function community service center (>2000 m²) with potential coordination with nearby facilities; and (3) moderate population size with varying aging levels to enable cross-case comparison.

2.2.2. Data Collection and Network Construction

Older adults’ routine facility use was recorded with a participatory, PPGIS-assisted behavioral mapping protocol, which improves spatial precision and interpretability in participatory settings (Kahila-Tani et al., 2019) [39]. In each community, ~30 independently mobile older adults marked routine destinations (≥2 visits per week) and their ordered sequences. This small-sample, high-density design is commonly used to reveal stable, high-frequency spatial–behavioral patterns (Rohrbach et al., 2016; Hasanzadeh, 2022) [53,54]. Each community sample is treated as a cross-sectional demonstrative subset to validate the analytical workflow rather than to estimate population parameters. To reduce noise and sparsity, analyses were confined to facilities actually visited, and rank-based permutation tests were used instead of parametric estimators.

Facility points and consecutive links were digitized to build two undirected weighted networks (nodes = facilities actually used):

• Behavioral Network—Edges connect facilities co-occurring in the same travel path; edge weights represent co-occurrence frequency.
• Effective Facility Network—For each facility pair, the shortest walking-path distance $d_{ij}$ was computed. Edge weights follow a continuous distance–decay: $w_{ij}=\exp \left(-{\displaystyle \frac{d_{ij}}{\lambda }}\right)$, a standard specification in accessibility and spatial interaction modelling (Geurs & van Wee, 2004) [55]. To avoid spurious long-range ties, weights beyond a maximum walking threshold $D_{\mathrm{m}\mathrm{a}\mathrm{x}}$ were set to zero.
• Additionally, structural stability across 500/800/1000 m was evaluated using matrix Spearman–QAP. Facility networks showed high cross-threshold consistency (Appendix A), indicating that structural judgments and core identification are not materially altered by the bounds. Given empirical considerations (Xu et al., 2024) [56], 800 m was adopted as the main specification because it retains sufficient connectivity while excluding weak, long-range ties.

2.2.3. Core Node Identification Using SNA

Social Network Analysis (SNA) was applied to assess structural characteristics and identify key facilities. The definitions of network structure and node centrality metrics follow established works in social network analysis (Wasserman & Faust, 1994; Freeman, 2002; Opsahl et al., 2010) [57,58,59]. Calculations were performed in Python 3.13.5 using the NetworkX 3.4.2 package. Analyses employed undirected, weighted graphs with self-loops removed.

(1)

Behavioral Network Structure Metrics

• Weighted Density: Indicates the overall compactness of behavioral linkages among facilities.
• Weighted Clustering Coefficient: Measures the extent of local triadic closure, revealing co-use clusters among facilities.
• Centralization Index: Evaluates how strongly the network is organized around a few dominant nodes. It was computed from weighted degree scores and normalized to [0, 1]; higher values reflect concentration of activity chains, while lower values imply a balanced structure.
• Normalized HHI: Captures the concentration of behavioral activity across facilities based on their usage frequencies. The index was normalized as $(\mathrm{HHI}-1/n)/(1-1/n)$ to allow cross-community comparison. Higher values indicate behavioral dependence on a few facilities, while lower values reflect balanced use.

(2)

Node Centrality Metrics (Weighted Forms)

• Degree Centrality: Represents direct connectivity and usage frequency of each facility.
• Eigenvector Centrality: Identifies facilities connected to other highly used nodes, highlighting their organizational potential.
• Closeness Centrality: Indicates accessibility and efficiency in reaching other facilities.
• Betweenness Centrality: Reflects intermediary roles along behavioral routes; due to boundary sensitivity at the community scale, it is used only as a supplementary metric.

2.2.4. Structure–Behavior Coordination

Coordination was assessed at the node level with two complementary measures:

(1)

Overall rank consistency—Rank-QAP.

For each centrality metric, ranks from the facility and behavioral networks were compared using Spearman’s $\rho$. Statistical significance was obtained via the Quadratic Assignment Procedure (node-label permutations) (Krackhardt, 1987; Dekker et al., 2007) [60,61], which accounts for network autocorrelation and yields more robust inference than conventional Spearman tests. Larger Rank-QAP $\rho$ indicates stronger global agreement between structural and behavioral importance.

(2)

Head alignment—Rank-Biased Overlap (RBO).

Agreement among the most prominent nodes was evaluated with RBO, which applies geometrically decaying weights to lower ranks (Webber et al., 2010) [62]. The depth parameter was set size-adaptively (≈ top ~20% of nodes), and significance was obtained via node-label permutations. Compared with a fixed Top-$k$ overlap, RBO is continuous, less threshold-sensitive, and captures both membership and ordering among leading nodes; larger RBO denotes stronger front-end coordination.

3. Results

3.1. Spatial Structure of Three Representative Cases

3.1.1. Aggregation Patterns and Case Selection

As described in Section 2.2.1, spatial analysis of facility distribution identified three distinct aggregation patterns across Guangzhou (Figure 2):

• Clustered Type: Concentrated in small plots and open blocks of the old city, with compactly mixed service functions, broad catchment areas, and high facility density.
• Linear Type: Belt-shaped clusters mainly along living streets, often located at the edges of gated residential areas, shaped by both planning policy and market forces.
• Patchy Type: Composed of multiple high-density nodes or complexes with relatively dispersed layouts but strong functional integration, often formed through transit-oriented development (TOD) or commercial agglomeration.

Based on these patterns, three representative communities were selected: FY (clustered), DJ (linear), and LG (patchy). FY and DJ have higher aging shares (~25% and ~20% aged 60+), indicating strong demand for age-friendly services; LG has a lower aging rate (~7%) but represents a newly developed affordable-housing area with weaker surrounding services, offering a contrasting context for accessibility and activation in patchy structures.

3.1.2. Community Profiles

Figure 3 shows the spatial contexts and service center locations of the three selected communities. Facility provision was evaluated from the perspective of each community service center using three indicators—facility density, information entropy, and mixed-use degree—within 300 m, 500 m, and 800 m radii (

Table 1. Facility indicators of three communities at different service radii.

Community	Facility Density (Per km2)			Information Entropy			Mixed-Use Degree
	300 m	500 m	800 m	300 m	500 m	800 m	300 m	500 m	800 m
FY	516.95	775.58	828.98	1.361	1.576	1.632	0.577	0.652	0.682
DJ	507.43	348.31	261	1.232	1.348	1.552	0.516	0.557	0.624
LG	1178.27	540.93	323.15	1.626	1.667	1.684	0.672	0.682	0.689

Coloring rule: Values in each column are color-coded; red indicates higher values, blue indicates lower values.

).

(1)

FY Community–Clustered Aggregation

Located in the old city, FY features low-rise Li-fang housing and a dense street–alley network typical of traditional Lingnan neighborhoods. Facilities are tightly clustered along pedestrian streets, with the service center embedded at the core, ensuring strong accessibility and spatial continuity. The indicators show consistently high density and mixed use across all radii, reflecting compact structure and extensive service coverage.

(2)

DJ Community–Linear Aggregation

DJ lies on the urban fringe, composed of high-rise commercial and affordable housing. Facilities align along Dongjiao South Road, forming a belt-shaped corridor. The service center, located on the top floor of a commercial complex, suffers from weak vertical connectivity. Density decreases rapidly with distance, while entropy and mix increase, suggesting a dispersed yet functionally balanced layout.

(3)

LG Community–Patchy Aggregation

LG, a suburban affordable-housing area, exhibits uneven facility provision. Facilities cluster around the Yehoo Fong complex, with the service center positioned adjacent as a key public anchor. High density within 300 m and sharp decline beyond indicate a patchy pattern with a concentrated core and sparse periphery.

In summary, the three communities exhibit clear differences in facility spatial structure, service center embedding, and surrounding functional mix, forming a solid empirical basis for examining how spatial organization affects older adults’ service experiences and structure–behavior coordination.

3.1.3. Demographic Characteristics

To contextualize the behavioral mapping, we report the demographic profile of participants across the three communities (n = 30 per community). The samples cover both genders, three age bands (60–69, 70–79, 80+), and multiple income brackets, supporting within- and cross-case comparisons despite the limited size. Detailed counts are provided in

Table 2. Demographic characteristics of older adult participants in the three case communities.

Community		FY	DJ	LG
Gender	Male	13	11	15
	Female	17	19	15
Age	60–69	10	13	18
	70–79	17	15	12
	80+	3	2	0
Household Per Capita Annual Income (CNY)	Below 30,000	13	8	18
	30,000–50,000	15	9	9
	50,000–100,000	2	8	3
	100,000–200,000	1	4	0
	200,000 and above	0	1	0

.

3.2. Overall Structural Characteristics of Behavioral Networks

Table 3. Structural metrics of behavioral networks in the three communities.

Community	Number of Nodes	Number of Edges	Weighted Network Density	Weighted Average Clustering Coefficient	Centralization Index	Normalized HHI
FY	28	201	1.323	0.088	2.95%	4.16%
DJ	30	194	0.83	0.113	2.61%	2.19%
LG	17	82	1.316	0.089	5.14%	4.52%

summarizes the structural metrics of behavioral networks for the three case communities. These results are used to compare network compactness, clustering tendencies, and behavioral concentration across different facility aggregation forms.

FY shows the most compact and balanced behavioral structure, with the highest weighted density (1.323) and moderate centralization (2.95%) and HHI (4.16%). This indicates a well-connected network where older adults’ activities are broadly distributed without overreliance on specific facilities.

DJ contains the largest number of nodes (n = 30) but has weaker overall connectivity (density = 0.83). Its clustering coefficient (0.113) is the highest, suggesting locally cohesive but more dispersed behavioral linkages.

LG, with the smallest network (n = 17), demonstrates high centralization (5.14%) and elevated HHI (4.52%) despite comparable density (1.316), implying that behavioral flows are concentrated around a few key facilities and exhibit stronger path dependence.

Overall, the three communities differ substantially in network structure: FY’s compact and evenly connected system supports diverse activity chains, DJ’s dispersed layout shows fragmented but localized interactions, and LG’s concentrated structure reflects reliance on limited behavioral hubs.

3.3. Identification of Core Facility Nodes and Analysis of Behavioral Centers

Facility- and behavior-network metrics were computed for all nodes (full rankings in Appendix B). Figure 4 visualizes behavioral mapping: node size encodes usage frequency; edge thickness encodes co-occurrence along consecutive activity chains; Node No. 1 denotes the community service center; red-filled nodes indicate neighborhood cores. We define neighborhood cores as facilities showing a clear performance gap in the behavioral network and usage frequency ≥ 15 of 30 respondents (≥50%). Accordingly, cores within 100 m of the service center are classified as compact; those beyond 100 m are classed as extended (

Table 4. Characteristics of neighborhood cores in three communities.

Community	Neighborhood Core Combination	Functional Type	Core Type	Institutional Platform Embedding
FY	FY1 (Service Center), FY2 (Pocket Park), FY5 (Fitness Plaza)	Community Service + Open Space	Compact	Embedded
DJ	DJ2 (Commercial Complex), DJ20 (Community Green Space)	Commercial Complex + Open Space	Extended	Weakly Embedded
LG	LG1 (Service Center), LG2 (Commercial Complex)	Community Service + Commercial Complex	Compact	Embedded

).

FY: Older adults’ activity paths are highly concentrated within 500 m of the service center. FY1 (Service Center), FY2 (Pocket Park), and FY5 (Fitness Plaza) constitute the dominant hub, each scoring > 0.7 in degree centrality in both the facility and behavioral networks. Peripheral nodes, such as FY24 (Hengbao Plaza) and FY16 (Liwan Lake Park), exhibit lower structural centrality but still appear frequently in activity chains, indicating supplementary use beyond the compact core.

DJ: Activity paths are distributed along an approximately 800 meter stretch of Dongjiao South Road. DJ2 (Fanghehui Complex) and DJ20 (Community Green Space) anchor the behavioral network as the most frequently visited facilities. The service center (DJ1), despite its proximity to DJ2, ranks lower in both networks. In contrast, several open-space nodes—such as DJ26 (Fangcun Garden Green) and DJ27 (Basketball Court)—exhibit high behavioral centrality despite their low structural prominence, underscoring the importance of accessible outdoor spaces for older adults.

LG: Behavioral activities cluster within a 500 m radius around LG1 (Service Center) and LG2 (Yehoo Fong Complex). Both nodes rank high in centrality across the two networks, forming a clear dual-core structure. Meanwhile, LG3 (Bus Stop) and LG5 (Metro Station) function as secondary connectors that link peripheral areas. This configuration reveals a smaller yet more cohesive network, which is dominated by multifunctional nodes that integrate community services with everyday amenities.

3.4. Responsiveness Analysis of Facility and Behavioral Networks

Node-level coordination was assessed by Rank-QAP and RBO (chosen to approximate the top ~20% for each community: FY $p=0.821$, DJ $p=0.833$, LG $p=0.706$) (

Table 5. Rank-QAP and RBO (permutation p-values).

	Metric	Spearman ρ (Rank-QAP)	Spearman p (Perm)	RBO (p = Adaptive)	RBO p (Perm)
FY	degree	0.335	0.0862	0.323	0.0976
	eigenvector	0.35	0.0708	0.408 *	0.046
	closeness	0.361	0.062	0.301	0.1264
DJ	degree	0.064	0.7357	0.123	0.759
	eigenvector	0.056	0.7676	0.13	0.7259
	closeness	0.013	0.9486	0.143	0.6443
LG	degree	0.396	0.1198	0.323	0.133
	eigenvector	0.17	0.5133	0.288	0.1874
	closeness	0.697 *	0.002	0.547	0.059

* indicates statistical significance at the 0.05 level. FY: eigenvector RBO indicates significant head alignment (RBO = 0.408, $p<0.05$); overall rank consistency is weak–moderate and near-significant (ρ = 0.350, $p=0.071$). DJ: no significant consistency across degree/eigenvector/closeness by either Rank-QAP or RBO. LG: closeness shows strong overall rank consistency (ρ = 0.697, $p=0.002$) and borderline head alignment (RBO = 0.547, $p=0.059$). These patterns indicate heterogeneous dominance of consistency dimensions across communities (prestige-led head alignment in FY; accessibility-led global alignment in LG; weak alignment in DJ).

). Results focus on significant or near-significant findings:

4. Discussion

4.1. Compatibility of Community Service Centers Across Spatial Aggregation Forms

The spatial aggregation of facilities shapes the operating environment of community service centers and the conditions in which older adults navigate and use them. Understanding these contextual effects is crucial for explaining how spatial potential translates into everyday behavior.

In the clustered case (FY), facilities are dense and functionally mixed, producing a compact yet balanced behavioral structure. The service center occupies an axial intersection with adjacent open spaces, benefiting from strong visibility and direct pedestrian flows. These findings echo Mouratidis and Poortinga’s (2020) [63] observation that compact, mixed-use, and legible urban forms foster vitality and social cohesion by translating spatial potential into everyday encounters. Coordination appears as head-only alignment on prestige-type nodes, a small set of highly connected facilities jointly anchor use, while agreement does not extend across the full ranking. Planning should strengthen legibility and accessibility around these prestige nodes, compound hub functions, and implement fine-grained barrier-free upgrades to sustain mobility.

In the linear case (DJ), facilities align along a corridor with segmented connectivity. The service center, located on upper floors of a commercial complex, suffers from limited ground-level exposure and weak behavioral translation, whereas the street-facing mall and green space act as active cores. Studies on street network morphology show that both physical and visual connectivity significantly influence pedestrian flows, and disrupted connectivity reduces foot traffic even in areas with high facility density (Hajrasouliha & Yin, 2015) [64]. Neither global rank consistency nor head alignment is significant, showing that proximity alone cannot induce behavioral convergence. Planning should close network breaks, enhance wayfinding, and reprogram latent nodes; entrances and interfaces of service centers should align with main pedestrian corridors to convert structural integration into actual use.

In the patchy case (LG), the network relies on a few high-accessibility nodes forming a dual core of service center and commercial complex, complemented by transit stops that connect peripheral links. Coordination manifests as accessibility-driven alignment, where continuous paths and short walking distances enable behavioral cohesion across the network. Consistent with Yang et al. (2022) [65], micro-scale connectivity improvements and secondary access routes yield disproportionate gains in local accessibility, helping dispersed community nodes function as cohesive behavioral systems. Planning should emphasize walkability and micro-connections rather than magnifying a few prestige nodes; pairing the service center with commercial functions can enhance visibility and offset weak peripheral provision, improving the conversion from structural potential to sustained use.

Across these cases, differences in coordination reveal how spatial form conditions behavioral outcomes: compact and visible configurations convert structural potential into actual use more efficiently, while discontinuous or vertically separated layouts require deliberate spatial and programmatic mediation. These variations highlight that spatial potential alone is insufficient; effective coordination depends on accessibility, legibility, and anchor performance, which collectively mediate the translation from built form to everyday behavior (Ewing & Handy, 2009; Dovey & Pafka, 2020) [66,67]. These findings are consistent with broader empirical evidence. As Li and Li (2023) [68] note, facilities along arterial streets or within composite developments attract higher footfall, demonstrating that openness and accessibility are decisive for sustained use. Residents—particularly older adults—prefer composite, open community centers that integrate everyday services with social and recreational functions (Yu & Zhu, 2024) [69]. These spatial principles are exemplified in planning practices that institutionalize accessibility and mixed-use integration. Singapore’s neighborhood complexes offer a relevant precedent: maintaining a balanced commercial share (40–60%) not only supports financial viability but also enables intergenerational interaction and one-stop accessibility (Zhao et al., 2024; Heng, 2017) [70,71]. Similar mechanisms are observed in European contexts, where multi-functional community hubs enhance daily mobility and reduce social isolation (Buffel & Phillipson, 2018) [72].

Quantitative evidence from Rank-QAP and RBO further reinforces this interpretation: alignment between facility potential and observed behavior varies by community and metric. Only LG exhibits significant closeness-based correlation, indicating that spatial accessibility can drive network-wide behavioral alignment. Hence, spatial proximity is necessary but not sufficient for interaction. True integration arises when structural centrality converges with behavioral routes through legible, continuous, and socially vibrant interfaces (Yang et al., 2022) [73]. Service centers thus become genuine behavioral hubs only when spatial accessibility and everyday movement coincide, transforming administrative platforms into active neighborhood cores that sustain aging in place.

4.2. Core Nodes and Anchor Mechanisms: Structural Foundations of Older-Adult Service Networks

Regardless of whether a network is concentrated or decentralized, high-centrality nodes anchor older adults’ activity paths and form the backbone of community service networks. Since daily travel often involves leisure, shopping, household chores, and intergenerational care, leisure and household activities dominate (Hu et al., 2013) [74], producing repetitive high-frequency chains. Nodes that repeatedly appear in these chains typically combine multiple functions, strong accessibility, and open interfaces, constituting the main chain of the community network. In our mapping, wet markets and supermarkets (e.g., FY17, DJ3, LG6) emerged as core nodes due to high-frequency consumption, while parks (e.g., FY2, DJ20, DJ26) attracted substantial leisure use. Even destinations beyond 800 m, such as FY16, remained popular for their social and environmental value, supporting evidence that parks and green spaces promote community activity among older adults (Wang et al., 2022; Eronen et al., 2014) [75,76]. Some city-scale facilities, though peripheral and less connected in the facility network, possess highly composite programs, quality environments, and strong spatial legibility. They exert powerful behavioral pull and act as anchors of the community life circle. For example, FY16 saw Liwan Lake Park and DJ5 Lisheng Plaza—large green spaces and commercial complexes, respectively—draw activity flows and expand behavioral boundaries. Such spaces exemplify a “mix of mixes” (Dovey & Pafka, 2020) [67], shaping both spatial structure and behavioral organization. In dense old-city contexts, anchors can substitute for missing internal functions, thereby increasing the resilience and completeness of the overall service network.

Formation logics differ by settlement history. In legacy FY, hubs cluster in small everyday public spaces (wet markets, pocket parks, street plazas) shaped by long-term self-organization. In newer DJ and LG, behavioral paths rely more on planner-led neighborhood complexes, producing a planning-driven pattern. LG shows a concentrated dual core with sparse peripheries and repeated reliance on medium-to-large nodes; public transport provides a crucial intermediary that sustains reach and health-related mobility, especially where provision is thin or spatially dispersed (Franke et al., 2017; Guo et al., 2024) [19,77].

In sum, life-circle convenience depends on system-wide coordination rather than single facilities. Well-positioned community centers can enhance public facility efficiency and catalyze surrounding markets to diversify services (Ruan et al., 2021) [78]. As institutional platforms, centers should co-locate with high-frequency facilities and integrate behavior-chain-oriented networks, prioritizing markets, parks, and sports venues closely tied to daily life. In space-constrained old districts, large parks can serve as anchors to offset deficits in internal public space. Future development and renewal should strengthen functional linkages between centers and anchors, ensuring basic provision while enhancing social interaction and health support to form adaptive, age-friendly networks.

4.3. Differentiated Optimization Strategies Based on Structure–Behavior Coordination

The structure–strong/weak × behavior–strong/weak quadrant (Figure 5) diagnoses gaps between spatial potential and observed use, showing how facilities in each community manifest distinct coordination patterns. It supports differentiated design and management responses, recognizing that built form and everyday practice are co-produced in neighborhood settings.

(1)

Structure–strong × behavior–strong:

These nodes (e.g., FY1, LG1) sit centrally in both networks and sustain high co-use. Design responses should prioritize maintaining accessibility and comfort over large-scale transformation: continuous barrier-free routes, shade/seating, lighting, and clear wayfinding, complemented by low-cost programming (e.g., volunteer fairs, pop-up markets). Improving walkability and access to nearby blue–green spaces can extend hub influence, integrate adjacent nodes into continuous chains, and support older adults’ well-being (Shen et al., 2025) [79].

(2)

Structure–weak × behavior–strong:

Despite modest structural centrality, open, visible, socially attractive sites draw heavy use (e.g., DJ26/DJ27 in the linear corridor; FY24 in the clustered layout). Incremental upgrades that formalize comfort and strengthen network integration—shaded seating, lighting, pavement repair, and micro-connectivity such as side-path shortcuts or mid-block crossings—are appropriate. Where possible, light institutional support or small-scale service consolidation at edges can evolve these sites into micro-activity hubs without constraining existing spontaneity.

(3)

Structure–weak × behavior–weak:

Facilities in this quadrant, often dispersed residuals, call for rational consolidation rather than capital-intensive upgrades. Repurposing them for auxiliary community functions or merging with nearby active hubs prevents resource fragmentation and improves spatial efficiency.

(4)

Structure–strong × behavior–weak:

Here, spatial prominence fails to translate into routine use due to broken paths, limited frontage openness, or narrow functional scope (e.g., LG7–LG9). Functional mismatch should be distinguished from under-activated potential: the former requires transformation or reallocation, whereas the latter can be addressed by improving accessibility (continuous pedestrian links, ground-level entries) and synergy with adjacent high-frequency facilities. High-density communities often exhibit spatial abundance but behavioral underuse, a pattern Wang et al. (2025) [80] attribute to low perceived accessibility and limited legibility despite physical proximity. Enhancing visibility and ground-level access can thus reactivate underperforming nodes without structural expansion. These targeted retrofits maximize existing assets at low cost, consistent with the “light and precise” approach in adaptive community renewal.

Building on prior work that shows context-specific, multi-factor interventions outperform single-factor measures for ageing in place (E et al., 2024) [81], this study operationalizes a quadrant-based strategy that tailors actions—path optimization, functional transformation, and accessibility enhancement—to node attributes. Under constrained resources, prioritization benefits from a dual, cross-community lens. From a public-asset perspective, structure-strong × behavior-weak nodes should be upgraded first to improve returns on existing investments and strengthen network connectivity. From a community-renewal perspective, quick, low-cost gains can be achieved by refining structure-weak × behavior-strong nodes, where modest comfort and micro-connectivity fixes amplify evident behavioral energy. Running these tracks in parallel—short-term activation of vibrant places alongside progressive retrofitting of underperforming but structurally salient facilities—offers a staged, cost-effective pathway to evolve institutional nodes into behavioral hubs while supporting ageing in place.

5. Conclusions

This paper positions structure–behavior coordination as a diagnostic, facility-centric and sequence-aware framework for examining how the spatial configuration of age-friendly community facilities aligns with older adults’ observed activity-chain use. By coupling PPGIS-based behavioral mapping with social network analysis, it provides an operational model and comparable metrics for assessing alignment between facility networks and behavioral networks.

Methodologically, the research establishes a replicable analytical pipeline integrating node-level Rank-QAP (overall rank consistency) and RBO (head alignment) metrics. This sequence-aware, permutation-based approach reduces reliance on large continuous-tracking datasets and is portable across community types and cities. From a planning perspective, the findings highlight that coordination is both relational and context-dependent. Different communities exhibit distinct dominant coordination dimensions—some governed by structural prestige, others by accessibility and continuity—suggesting that age-friendly planning should prioritize context-specific integration rather than universal metrics. The structure–behavior typology thus serves not as a prescriptive model, but as a flexible diagnostic tool to guide resource allocation, spatial retrofitting, and behavioral activation under constrained conditions.

Findings stem from three Guangzhou cases and a cross-sectional mapping of independently mobile older adults; they illustrate the framework’s utility rather than claim universal effects. Broader validation should test additional cities and morphologies, incorporate multi-season and event windows, and triangulate PPGIS with GPS traces and interviews/time-diaries to enrich chain inference. Extending the protocol to varied functional-ability groups will help generalize toward a whole-age-friendly approach. Despite these bounds, the proposed framework—combining usage-anchored core identification, distance-decay facility networks, and complementary rank-based alignment tests—offers a transferable method to diagnose, compare, and act on structure–behavior coordination in age-friendly community systems.