Dynamic Measurement and Equity Analysis of Walking Accessibility in Primary Healthcare Institutions Under Diverse Supply–Demand Scenarios: Evidence from Shenyang

Abstract

The walking accessibility of primary healthcare institutions (PHCIs) is a pivotal determinant of health equity. However, prior studies often lack a comprehensive assessment that integrates the spatiotemporal dynamics of both multi-faceted supply and multi-scenario demand. To bridge this gap, this study develops an enhanced two-step floating catchment area method (2SFCA-MSD) that concurrently incorporates multiple types of service supply and multiple temporal demand scenarios to quantify PHCI walking accessibility, with equity evaluated using the Gini coefficient and Lorenz curve. The results indicate that: (1) Both supply and demand exhibit pronounced spatiotemporal inequalities. (2) Walking accessibility varies substantially across scenarios; Health services for vulnerable groups (Service B) exhibit the highest walking accessibility across all three supply scenarios, while the morning work scenario demonstrates the best walking accessibility among the four demand scenarios. (3) Gini coefficients exceeding 0.5 across all scenarios reveal severe resource allocation inequity. By establishing a dynamic supply–demand integration framework, this research advances methodological precision in accessibility evaluation, uncovers critical spatiotemporal mismatch patterns, and provides actionable insights for optimizing PHCI planning to promote spatial justice in urban health.

1. Introduction

Primary Healthcare Institutions (PHCIs) refer to the first point of contact for individuals within a healthcare system, typically including community health centers, clinics, and general practitioner services, which provide comprehensive, accessible, and continuous care. As critical infrastructure for improving population health equity and bridging healthcare system disparities, PHCIs have garnered increasing attention from international organizations and the academic community. Walking accessibility refers to the ease with which residents can reach PHCIs on foot within a given time threshold, reflecting the interaction between spatial location and mobility constraints [1]. In contrast, walkability describes the built environment’s pedestrian-friendly attributes—such as sidewalk connectivity, safety, and land-use mix—that influence the feasibility and attractiveness of walking [2]. The World Health Organization (WHO) projects that expanding primary healthcare interventions in low- and middle-income countries could save tens of millions of lives and significantly increase average life expectancy by 2030 [3]. From the Alma-Ata Declaration to the Astana Declaration, global consensus has consistently emphasized that primary healthcare is the cornerstone for achieving universal health coverage [4,5]. Empirical research further demonstrates that a robust primary healthcare system is effective in managing chronic diseases, responding to public health crises, and enhancing community resilience [6,7,8].

The methodological evolution in quantifying healthcare facility accessibility profoundly reflects the continuous deepening of the understanding of “spatial equity” within the disciplines of geography, public health, and urban planning [9]. This developmental trajectory—progressing from early, coarse macro-statistics, to models simulating spatial interactions, and further to current frontier explorations pursuing multi-dimensional and fine-grained characterizations—represents not merely an improvement in computational techniques, but a sustained revelation of the complex connotations embedded within “accessibility,” including spatial impedance, supply–demand matching, and individual equity.

In the initial stages of accessibility research, methodologies primarily relied on supply-oriented ratio methods and minimum distance approaches. The former, exemplified by metrics like “number of beds per thousand population,” facilitated macro-level matching of aggregate resources at a regional planning scale [10]. The latter calculated the straight-line (Euclidean) distance from a demand point to its nearest facility [11,12]. These methods, characterized by intuitive principles and simple data requirements, remain in use for certain macro-policy evaluations today. However, their limitations are fundamental and severe: they entirely sever the spatial interconnection between supply and demand. Ratio methods mask substantial intra-regional disparities, as the highly concentrated resources in one part of a region and extreme scarcity in surrounding areas can be misleadingly presented by an averaged “per capita” figure. The minimum distance approach overlooks the service capacity of supply points and competitive effects, treating a clinic with a single doctor as equivalent to a major hospital in its ability to serve the surrounding population. In essence, these approaches constitute a form of “static distribution description” [13] rather than a genuine measurement of accessibility, failing to address the core equity question: “Can residents at a specific location obtain adequate services at an acceptable cost?”.

To overcome the aforementioned limitations, the gravity model from physics and its significant adaptation in public health—the Two-Step Floating Catchment Area (2SFCA) method—were introduced [1,14,15]. This paradigm revolutionized the conceptualization of accessibility by defining it as the spatial interaction opportunity between supply and demand. Its core principle incorporates a distance decay function, acknowledging and quantifying two fundamental realities: first, that the propensity for residents to utilize a healthcare service declines with increasing travel cost [1], and second, that a facility’s service capacity is distributed among nearby populations, with its effective “pressure” or influence diminishing over distance [16]. Represented by the 2SFCA method and its enhanced variants (e.g., E2SFCA [17], 3SFCA [18]), these spatial interaction models rapidly became the dominant mainstream methodology over the past fifteen years. Their adoption is widespread, owing to the logical framework and interpretable results they provide for assessing the equity of healthcare resources across scales ranging from cities to nations.

However, recent research indicates that significant limitations persist across three dimensions in accurately characterizing the accessibility of PHCIs, which forms the starting point of this study. First, on the supply side, there is a prevalent tendency to treat facilities as homogeneous entities, overlooking internal variations in service types and capacities. Most studies aggregate healthcare institutions into a single unit when calculating supply capacity, thereby masking the heterogeneity of their internal service structures. For instance, resident demand for specific services such as preventive care, chronic disease management, or pediatrics may not be adequately captured by monolithic indicators like “number of physicians” or “bed count.” Zeng et al. explicitly state that neglecting the typological diversity of medical facilities leads to an underestimation of both accessibility and equity [19]. In other words, an area might exhibit high “aggregate accessibility” due to the presence of a large general hospital, while simultaneously experiencing a scarcity of the most urgently needed basic medical services at the community level.

Second, on the demand side, reliance on static residential population data fails to capture the spatiotemporal dynamics of demand. Canonical models typically use nighttime population distribution as a proxy for demand, which cannot reflect the drastic spatial reconfiguration of the population during daytime hours due to activities such as living, working, and leisure. This static assumption in demand modeling, by ignoring spatiotemporal dynamics, renders accessibility assessments detached from real-world contexts and unable to guide time-sensitive and refined resource allocation.

Third, regarding transportation mode, there is insufficient fine-grained modeling of walking, the fundamental mode of access for PHCIs. For PHCIs primarily accessed by walking, the precision offered by simple Euclidean distance or even standard network analysis is often inadequate. The actual walking experience is significantly influenced by factors such as street network connectivity and intersection delays. For example, a study employing Ga2SFCA (Gaussian-enhanced 2SFCA) integrated with walking impedance and network analysis to assess the spatial resilience and service coverage of primary healthcare facilities offers a new perspective for layout optimization [20]. This demonstrates that constructing a fine-grained walking time-cost model is crucial for evaluating PHCIs, particularly in high-density urban built-up areas.

Therefore, the core innovative positioning of this study lies in its commitment to constructing a coupled assessment framework—termed the 2SFCA-MSD (Multi-Supply and Multi-Scenario Demand enhanced 2SFCA)—that simultaneously integrates “multi-type service supply,” “multi-spatiotemporal scenario demand,” and “fine-grained walking time cost.” This framework is designed to achieve three primary objectives:

• Deconstruct supply heterogeneity: Differentiate and quantify the supply capacity of various basic medical services within individual PHCIs.
• Simulate demand dynamism: Utilize multi-source spatiotemporal big data to model dynamic population demand across different temporal cross-sections.
• Model fine-grained walking time cost: Calculate realistic walking times that incorporate multiple impedance factors (e.g., intersection delays, pedestrian infrastructure), based on actual navigation time data from the Gaode Map API, rather than relying on idealized network distances.

Through this systematic integration, this study aims to overcome the bottlenecks of existing methods in depicting complex realities. It seeks to provide a more refined, dynamic, and realistic map of PHCI accessibility and equity that closely aligns with residents’ actual healthcare-seeking experiences. Consequently, it aspires to offer scientifically robust and reliable decision-making support for implementing proactive and precise planning of primary healthcare resources in both urban and rural settings. This endeavor represents not merely a response to a methodological gap, but a profound methodological practice in the scientific measurement and enhancement of the spatiotemporal equity of essential public services.

2. Materials and Methods

2.1. Research Area Overview

Shenyang is situated in southern Northeast China, at the center of the Liaohe Plain, and features a temperate humid continental climate with four distinct seasons (Figure 1). Winters are cold and prolonged; the coldest month, January, has an average temperature of approximately −11 °C. As the provincial capital of Liaoning, Shenyang functions as an economic and cultural hub for the region. Its core development goals include becoming a national advanced manufacturing base, a modern service-industry center for Northeast China, and a regional science and technology innovation hub. In 2024, Shenyang’s regional gross domestic product (GDP) reached RMB 902.71 billion, with the three-sector ratio being 3.6:35.1:61.3. The per capita disposable income of residents was RMB 49,758 [21]. Rising income levels are generally associated with increased healthcare expenditure [22]. By the end of 2023, Shenyang had a permanent resident population of 9.204 million, of which 7.834 million (85.12%) lived in urban areas—well above the national urbanization rate of 66.16% [23]. The registered population aged 60 and above exceeded 2.24 million, accounting for 29.4% of the total [24]. According to United Nations aging criteria, Shenyang is approaching the threshold of a “severely aged society.” Population aging has driven a surge in demand for basic medical services [25]. As of late 2023, Shenyang contained 5193 primary healthcare institutions, including 113 township health centers, 152 community health service centers, 596 outpatient clinics, and 2004 village health clinics. Despite this substantial number of facilities, they remain insufficient to meet the massive annual demand of approximately 45.9 million patient visits.

The Second Ring Expressway was selected as the study boundary due to its clear spatial definition and socio-economic relevance. First, the area within the Second Ring constitutes a high-density residential zone and contains a concentration of medical service facilities in Shenyang, including 102 PHCIs with relatively comprehensive service functions. Second, the road network inside this area follows a “ring-and-radial” pattern, with the inner secondary roads often experiencing severe congestion. This discrepancy between residents’ actual travel routes and theoretical service radii necessitates the use of electronic map data to account for real-time traffic conditions and assess true accessibility. Third, a distinct gradient exists in built environment and population structure between the areas inside and outside the Second Ring Road. By focusing on the inner area and excluding the interference of low-density suburban zones, this study can more precisely identify the “last-mile” problems in PHCI layout within high-density urban settings, thereby offering targeted planning strategies. The study area is presented in Figure 1.

2.2. Data Source and Processing

This study employed multi-source network data, with details summarized in Table 1 and elaborated below.

Data 1: Information on the names, geographic coordinates, and service profiles of all Primary Healthcare Institutions (PHCIs) within the study area was obtained from the official website of the Shenyang Health Commission. The website publishes a full list of institutions responsible for delivering basic public health services in Shenyang (https://wjw.shenyang.gov.cn/zwgk/fdzdgknr/jcws/202208/t20220803_3798985.html (accessed on 10 April 2024)). A total of 102 PHCI facilities were identified and geocoded into ArcGIS 10.8.2 using their latitude and longitude coordinates for spatial analysis (Figure 1).

Data 2: The population heatmap data were obtained from the Baidu Huiyan platform, a geographic big data service. These data are generated from aggregated, anonymized location information collected across Baidu’s mapping, search, and social application terminals, reflecting real-time population movement and density. Compared with traditional static census data, the heatmap offers a more dynamic representation of diurnal population changes and provides several practical advantages, including low acquisition cost, high spatial resolution, strong real-time performance, and robust privacy protection [26]. To systematically assess the resilience of primary healthcare services under extreme adverse conditions, this study selects demographic and meteorological data from January 2024 as the analytical cross-section. This period is chosen based on the following dual considerations:

• Supply-side perspective: Winter represents the annual peak in healthcare demand. In January, temperatures in the study area drop to their lowest annual levels. Extreme cold weather has been shown to significantly increase the incidence of common diseases such as cardiovascular and respiratory conditions by activating the sympathetic nervous system, elevating blood viscosity, and other physiological mechanisms, thereby driving residents’ demand for primary healthcare services to its yearly maximum [27,28,29].
• Demand-side perspective: Winter corresponds to the annual bottleneck in pedestrian accessibility. Weather conditions in January, including snowfall, low temperatures, and road icing, severely reduce the efficiency and safety of the pedestrian network, resulting in the greatest spatial resistance for residents seeking healthcare services on foot. Thus, January represents a “worst-case scenario” characterized by the simultaneous pressure of “peak healthcare demand” and “poorest walking conditions.” By focusing on this temporal cross-section, this study aims to evaluate the spatial equity gaps of PHCI under the highest stress conditions of the year. A custom web crawler was developed in Python 3.8.10 to retrieve population distribution data from the Baidu Maps Open Platform. Data were collected for the study area hourly from 1 to 31 January 2024, resulting in 744 time-interval samples. To represent typical daily rhythms, the average population values at 03:00, 09:00, 15:00, and 21:00 were calculated, corresponding to four lifestyle scenarios: sleeping, morning work, afternoon work, and night leisure. The obtained data were imported into ArcGIS 10.8.2 for georeferencing, projection transformation, clipping, and kernel-density analysis.

Data 3: Residential land AOI data. The study area was delineated using the Second Ring Road as its boundary, based on the fundamental land-use classification map developed by Gong et al. [30]. While healthcare needs can arise across various land-use types (e.g., commercial, recreational, and transportation), this study focuses on basic public medical demand, which is predominantly concentrated in residential areas. Therefore, the centroid of each residential land parcel was adopted as the origin of service demand. Population counts for 348 residential parcels were obtained by spatially aggregating the Baidu heatmap data within the boundaries of these parcels.

Data 4: Walking travel times for medical visits were calculated between all 348 residential blocks (origin) and 102 PHCI facilities (destination) using the Route Planning API (v2.0) of the Amap (Gaode) Open Platform. To ensure stable and representative traffic conditions, all routing requests were submitted on 20 November 2024 (a regular Wednesday), avoiding holidays, extreme weather, and major incident periods. The obtained travel times reflect typical operational road and pedestrian network conditions.

Table 1. Research data sources.

Data	Data Source	Data Address
Baidu Population Heat Map Data	Baidu Map Open Platform [31]	https://lbsyun.baidu.com/faq/api?title=webapi/ip-api (accessed on 10 April 2024)
PHCI coordinates	Shenyang Municipal Health and Wellness Committee Website [32]	https://wjw.sheyang.gov.cn/zwgk/fdzdgknr/jcws/202208/t20220803_3798985.html (accessed on 10 April 2024)
Residential AOI data	Gong P et al. [30]	Mapping essential urban land-use categories in China (EULUC-China): Preliminary results for 2018
Walking time cost data	Gaode Map Open Platform [33]	https://lbs.amap.com/api/weserice/guide/api/newroute (accessed on 10 April 2024)

Table 1. Research data sources.

Data	Data Source	Data Address
Baidu Population Heat Map Data	Baidu Map Open Platform [31]	https://lbsyun.baidu.com/faq/api?title=webapi/ip-api (accessed on 10 April 2024)
PHCI coordinates	Shenyang Municipal Health and Wellness Committee Website [32]	https://wjw.sheyang.gov.cn/zwgk/fdzdgknr/jcws/202208/t20220803_3798985.html (accessed on 10 April 2024)
Residential AOI data	Gong P et al. [30]	Mapping essential urban land-use categories in China (EULUC-China): Preliminary results for 2018
Walking time cost data	Gaode Map Open Platform [33]	https://lbs.amap.com/api/weserice/guide/api/newroute (accessed on 10 April 2024)

2.3. Research Methods

This study employed a multi-source data approach and a multi-step analytical framework to assess the walking accessibility of residential areas to Primary Healthcare Institutions (PHCIs) and to evaluate associated equity issues. The overall research framework is illustrated in Figure 2. First, population heatmap data were programmatically retrieved via Python from the Baidu Maps platform for January 2024. To reflect dynamic daily patterns, the hourly data were averaged to represent four distinct lifestyle scenarios: sleeping (03:00), morning work (09:00), afternoon work (15:00), and night leisure (21:00). These population values were then aggregated to the residential grid level for subsequent analysis. Second, the service capacity of PHCIs was evaluated using the Analytic Hierarchy Process (AHP). A comprehensive assessment system was constructed based on three key dimensions: health affairs management, healthcare for vulnerable populations, and the prevention and control of common diseases. Third, the walking accessibility from each residential plot to all PHCIs was calculated using an enhanced Two-step Floating Catchment Area method that considers multiple types of supply and demand (2SFCA-MSD).

Finally, to quantify the equity of the accessibility outcomes, Gini coefficients were computed for the different service types and across the four time scenarios.

2.3.1. Construction of a Medical Service Capacity Evaluation System Based on the Analytic Hierarchy Process (AHP)

The Analytic Hierarchy Process (AHP) is a well-established method commonly applied to address poorly structured problems involving qualitative factors [34]. It operates by constructing a hierarchical model that integrates both qualitative and quantitative criteria, thereby supporting systematic decision-making [35]. In this study, the weightings for the target layer, normative layer, and action layer were determined via AHP using the Yaahp 0.5.3 software. This study first established a hierarchical framework. Through three rounds of Delphi expert consultation, an indicator system was determined. The composition of the expert panel aimed to comprehensively cover core professional perspectives influencing the walking accessibility of primary healthcare facilities: Medical Domain Experts (8): Four experts specializing in medical management possess firsthand knowledge of institutional operations, resource allocation, service standards, and quality control. Their expertise ensures the evaluation system aligns with the core concept of healthcare service “supply capacity.” Additionally, four experts specializing in clinical medicine provide critical insights from frontline diagnostic and treatment practices regarding institutional technical capabilities and patient needs, enhancing the clinical relevance of evaluation metrics. Urban Planning Experts (4): Focusing on urban spatial structures, pedestrian transportation systems, and land use, they holistically consider the spatial distribution of regional healthcare resources. This helps to establish a link between institutional service capacity and spatial accessibility. The Kendall’s coefficient of concordance W = 0.82 (p < 0.01) indicates a high degree of consistency among expert opinions. Next, the judgment matrix is constructed. Using a 9-point semantic scale, pairwise comparisons are conducted among factors within the same level to construct the judgment matrix. The numbers in the judgment matrix represent the relative importance between two factors. Third, based on the judgment matrix, the relative weights of each factor on the criteria of the upper level were calculated. The weight vector represents the importance of each factor in decision-making. A consistency test was conducted on the judgment matrix to ensure the rationality of the judgments. If CR < 0.1, the judgment matrix was considered to have acceptable consistency. Finally, the composite weights of each level factor relative to the target level were calculated to support the final decision.

2.3.2. PHCI Walking Accessibility Analysis Based on 2SFCA-MSD

The Two-step Floating Catchment Area (2SFCA) method, first introduced by Radke and Mu, quantifies spatial accessibility based on opportunity accumulation within a defined threshold, where a higher value indicates better accessibility [36]. While the traditional 2SFCA accounts for competition among residents for limited resources, it overlooks the influence of facility service capacity on healthcare-seeking decisions, relies on overly coarse static demographic data that fails to capture temporal dynamics, and does not adequately model the decay of travel willingness with distance.

To address these limitations, this study introduces an enhanced model termed the Two-step Floating Catchment Area Method considering Multiple Supply and Demand (2SFCA-MSD). The key refinements are threefold.

First, the search radius is defined using realistic travel time. Instead of Euclidean distance, which overestimates accessibility, walking time retrieved via the Gaode Maps API is used. This incorporates real-world impediments like congestion and traffic signals. In line with the “Urban Residential Area Planning and Design Standards” (GB50180-2018) [37], which stipulate that a PHCI should be available within a 15 min living circle, the search radius threshold is set to 15 min.

Second, both supply attractiveness and distance decay are incorporated. The comprehensive service capacity of a PHCI, derived from our AHP-based model, defines its attractiveness. A Gaussian decay function is then integrated to model how accessibility diminishes with increasing travel time.

Third, healthcare demand is dynamically characterized across multiple scenarios. To reflect intraday variations, Baidu population heatmap data are used to represent dynamic demand within four lifestyle scenarios: sleep (03:00), morning work (09:00), afternoon work (15:00), and night leisure (21:00). The population within each residential area during these time windows is counted as the potential demand.

The specific steps are as follows:

(1)

First, calculate the supply–demand ratio of PHCI: with PHCI j as the center, set the time threshold to 15 min, search for all residential points k within the search radius, and calculate the supply–demand ratio as shown in Equation (1).

$${ R}_j={\displaystyle \frac{S_j}{\sum _{k\in \{d_{kj}\le d_0\}}\hspace{1em}{D}_k}}$$

(1)

Among these: R_j represents the supply–demand ratio of the jth PHCI within the search threshold; S_j represents the supply capacity of the jth PHCI, which is comprehensively determined using the AHP method in this paper; D_k denotes the demand scale at point k, represented by population size; d_{k j} denotes the cost of accessing healthcare services between points k and j, typically expressed in terms of distance or time. In this study, d_{k j} is characterized using travel time costs obtained via the Gaode Maps API, representing the time required for residents to reach healthcare facilities on foot; d_o denotes the search radius threshold.

(2)

The second step is to search for all medical points j within the search radius centered on residential point i, add up all the services R_j provided by all medical points, and obtain the accessibility $A_i^F$ at residential point i, as shown in Equation (2).

$$A_i^F=\sum _{j\in \left\{d_{ij}\le d_0\right\}}\hspace{1em}{R}_j$$

(2)

In the formula: $A_i^F$ is the accessibility of PHCI at location i; d_ij is the time cost of obtaining medical services between points i and j.

The model incorporates a Gaussian decay function to express the decline in residents’ demand for medical care as the distance from PHCI increases. The improved two-step moving search method is expressed in Formulas (3)–(5) as follows:

$$A_i^F=\sum _{j\in \left\{d_{ij}\le d_0\right\}}\hspace{1em}{\displaystyle \frac{S_j\times f\left(d_{ij}\right)}{\sum _{k\in \left\{d_{kj}\le d_0\right\}}\hspace{1em}{D}_{k}f\left(d_{ij}\right)}}$$

(3)

$$f(d_{ij})=\left\{\begin{array}{ll}g\left(d_{ij}\right)& d_{ij}⩽d_0\\ 0& d_{ij}>d_0\end{array} \right.$$

(4)

$$g\left(d_{ij}\right)={\displaystyle \frac{e^{-\frac{1}{2}\times (d_{ij}/d_o{)}^2}-e^{-\frac{1}{2}}}{1-e^{-\frac{1}{2}}}}$$

(5)

In the formula: f(d_ij) is the general form of the distance decay function; g(d_ij) is the distance decay function within the search radius d_o; and Formula (5) is the Gaussian distance decay function expression (Figure 3). To enhance the comparability of walking accessibility across different scenarios, the A value representing walking accessibility is normalized and converted into a decimal number within the range of (0, 1).

2.3.3. Gini Coefficient and Lorenz Curve

The spatial equity of PHCIs within the study area was assessed using the Lorenz curve and the Gini coefficient. These established metrics, commonly applied in economics to measure income inequality, are equally pertinent for evaluating the distributional fairness of public resources like healthcare [38]. The Lorenz curve graphically represents the cumulative proportion of a resource (e.g., healthcare accessibility) against the cumulative proportion of the population. A greater deviation of the curve from the line of perfect equality (the 45° line) indicates a higher degree of inequality. The Gini coefficient provides a single quantitative index derived from the Lorenz curve, ranging from 0 (perfect equality) to 1 (maximum inequality). Following conventional interpretation [39], values below 0.4 are generally considered acceptable, while a coefficient above 0.5 signals a state of severe disparity. In this study, the Gini coefficient is calculated for different medical service types and across various time scenarios to quantify the level of inequity in PHCI walking accessibility.

The calculation formula is as follows:

$$GEu=1-\sum _{k=1}^n\hspace{1em}\left(P_k-P_{k-1}\right)\left(C_k+C_{k-1}\right)$$

(6)

$$C_k={\displaystyle \frac{\sum _{i=1}^k\hspace{1em}{A}_i{r}_i}{\sum _{i=1}^n\hspace{1em}{A}_i{r}_i}}$$

(7)

In the formula, GE_u represents the PHCI service equity index for geographical unit u; n denotes the total number of communities within geographical unit u; k refers to the kth community in the sorted list of PHCI accessibility values from smallest to largest, where k = 1, 2, …, n; A_i denotes the PHCI accessibility value for community i; r represents the population of community i; C_k is the cumulative proportion of the product of PHCI accessibility and community population from community i to community k, with C₀ = 0 and C_n = 1; P_k is the cumulative proportion of community population from community i to community k, with P₀ = 0 and P_n = 1.

3. Results

3.1. Dynamic Supply and Demand Evaluation of PHCI

3.1.1. Multidimensional Supply Capacity

This study applied the Analytic Hierarchy Process (AHP) to assess the multi-dimensional service capacity of Primary Healthcare Institutions (PHCIs), establishing a comprehensive evaluation system. The system comprises a normative layer with three service categories: Service A: Health Affairs Management (weight: 0.1429), Service B: Healthcare for Vulnerable Populations (weight: 0.4286), and Service C: Common Disease Prevention and Control (weight: 0.4286). The influence weights differ substantially among these categories, with Service A being the least weighted. The action layer further breaks down into 15 specific healthcare services. Among these, Family Planning (A3) has the lowest weight (0.0097), while Type 2 Diabetes Prevention and Treatment (C2) has the highest (0.1446); detailed weights are provided in

Table 2. PHCI multi-service capacity evaluation system.

Target Level	Normative Level		Action Level
PHCI servicecapacity	(A) Health Services Management	0.1429	(A1) Health Archiving	0.0551
			(A2) Health Education	0.0263
			(A3) Family Planning	0.0097
			(A4) HIV/AIDS Prevention	0.0269
			(A5) Health Literacy Promotion	0.0248
	(B) Health services for vulnerable populations	0.4286	(B1) Vaccinations	0.1000
			(B2) Children aged 0–6 Health care	0.1000
			(B3) Maternal health	0.0923
			(B4) Elderly health services	0.1000
			(B5) Chinese Medicine Health Services	0.0362
	(C) Common Disease Prevention and Control	0.4286	(C1) Hypertension prevention and control	0.1161
			(C2) Type 2 diabetes prevention and treatment	0.1446
			(C3) Mental Disease Prevention and Control	0.0337
			(C4) Tuberculosis control	0.0859
			(C5) Early warning of public health emergencies	0.0482

.

The study area encompasses seven administrative districts: Heping, Shenhe, Tiexi, Huanggu, Dadong, Hunnan, and Yuhong. A total of 102 PHCIs were included, revealing a markedly uneven and clustered spatial distribution. Facilities are concentrated in the central urban core, particularly around the junction of Shenhe, Heping, and Huanggu districts. In contrast, peripheral areas—including Hunnan, eastern Dadong, eastern Shenhe, Yuhong District, and northern Huanggu—exhibit dispersed and low-density PHCI coverage. Notably, Hunnan District contains no PHCI facilities.

The service provision capacity of the 102 PHCIs was operationalized through a weighted aggregation of 15 specific medical services (Figure 4). For each PHCI, the availability of each service was recorded as a binary variable (1 if provided, 0 if not). These binary indicators were then aggregated into the three composite service types (A, B, and C) by calculating a weighted sum based on the predetermined weights presented in Table 2. The resulting aggregated supply values for each service type were subsequently used to generate the kernel density distributions shown in Figure 5, illustrating the spatial concentration of different healthcare services.

The spatial provision of the three guideline-level medical services exhibits distinct patterns. Service C demonstrates the highest spatial coverage, concentrated around PHCI locations. Services A and B show similar, more limited extents of coverage. The high-value areas for all three service types collectively form a multi-centered spatial layout. Areas with the highest service density are primarily located in northern Shenhe District. Additional clusters are found in Yuhong District, southern Huanggu District, and eastern Tiexi District. In contrast, several areas show limited service provision: these include the Hun River corridor—encompassing Hunnan District, southern Heping District, and eastern Shenhe District—as well as historically industrial zones such as northern Tiexi District and eastern Daxing District.

3.1.2. Dynamic Demand for PHCI

The demand for PHCIs varies dynamically across different lifestyle scenarios (Figure 6). During the sleep scenario, demand peaks around Shenyang Railway Station and Peace Square. Moderate demand areas form a multi-centered pattern around Shenyang North Station, Tiexi Square, People’s Square, and the Shenyang Municipal Library, with overall demand in Tiexi District remaining relatively high. In contrast, lower demand levels are observed in most of Daxing District, eastern Shenhe District, and Hunnan District. In the morning and afternoon work scenarios, the highest demand is concentrated near the Municipal Library and Shenyang Railway Station. Moderate demand areas extend around Shenyang North Station, Zhongshan Square, and Tiexi Square, while demand declines toward the peripheral regions. The night leisure scenario exhibits robust overall demand throughout the study area. Peak demand is again focused around Shenyang Railway Station and People’s Square. Compared to other scenarios, moderate demand areas are more extensive, showing notable increases around Zhongjie Commercial Street, alongside sustained activity near Tiexi Square, the Municipal Library, and Shenyang North Station.

3.2. Walking Accessibility Based on 2SFCA-MSD

The study area consists of 348 residential blocks. Of these, only 159 blocks (45.7%) lie within a 15 min walking catchment of the nearest PHCI, while 189 blocks (54.3%) are located beyond this threshold. Walking accessibility varies considerably across the three primary healthcare service types (Figure 7). For Service A, the area percentages classified into the highest, high, medium, low, and lowest accessibility zones are 1.34%, 1.34%, 11.27%, 51.46%, and 34.59%, respectively. The corresponding distribution for Service B is 1.06%, 0.74%, 8.75%, 33.20%, and 56.25%. For Service C, the proportions are 1.34%, 2.96%, 14.59%, 44.83%, and 36.28%.

Notable variations in accessibility are also evident across the four daily life scenarios (Figure 8). Under the sleep scenario, 11.93%, 5.44%, 19.39%, 35.60%, and 27.63% of the area fall into the highest, high, medium, low, and lowest accessibility zones, respectively; the combined share of low and lowest zones exceeds that of the high and highest zones. During the morning commute, the distribution is 6.44%, 11.16%, 19.32%, 37.19%, and 25.87%. For the afternoon work scenario, the figures are 11.24%, 9.87%, 23.78%, 33.30%, and 21.82%. Under the night leisure scenario, they are 11.93%, 5.43%, 16.56%, 36.28%, and 29.79%. Spatially, the highest and high accessibility zones are predominantly clustered in central Huanggu District, central Tiexi District, and the junction of Shenhe and Dadong Districts. Conversely, the lowest accessibility zones are mainly found in eastern Huanggu District, northern Dadong District, northern Tiexi District, and central Heping District. Accessibility for individual residential blocks also shifts with scenarios. For example, a residential area near Xita displays medium, highest, high, and medium accessibility under the sleep, morning commute, afternoon work, and night leisure scenarios, respectively.

3.3. Evaluation of the Equity of PHCI

The Gini coefficients for the accessibility of the three PHCI service types are 0.6358 (Service A), 0.6314 (Service B), and 0.6308 (Service C). All values substantially exceed the 0.5 threshold, indicating severe inequity in resource distribution. Although Service C shows a marginally lower coefficient, the disparity among the three services is minimal.

This pronounced inequality is reflected in their Lorenz curves (Figure 9a), which overlap considerably. Approximately 40% of the population resides in areas with no access to a PHCI within a 15 min walk, corresponding to 0% cumulative service coverage. The near-identical curvature of the three curves confirms that the level of unfairness is very similar across all service types.

When examined by time scenario, the Gini coefficients are 0.7068 (sleep), 0.5990 (morning work), 0.7100 (afternoon work), and 0.7092 (night leisure). All values far exceed the 0.5 benchmark, confirming widespread unfairness in basic healthcare access. The morning work scenario demonstrates relatively better (though still poor) equity.

The corresponding Lorenz curves (Figure 9b) show that about 35% of the population lacks 15 min walking access to a PHCI. The curves for the sleep, afternoon work, and night leisure scenarios almost entirely overlap, indicating equivalent equity levels. The slightly lower curvature of the morning work scenario confirms its relatively low Gini coefficient.

4. Discussion

This study evaluates the methodological approach adopted in this study for assessing supply and demand and reflects the profound reconsideration and targeted expansion of existing research paradigms [14,40]. Moving away from the traditional reliance on macro-level indicators such as bed capacity and healthcare worker numbers, this research recognizes an inherent mismatch between such metrics and the heterogeneous service functions of PHCIs, which prioritize prevention and basic medical care. Empirical findings indicate that these conventional indicators often fail to accurately capture the true service capacity of many PHCIs that operate without inpatient beds or with limited physical space yet deliver comprehensive services.

This methodological shift aligns with recent critical reflections within academia on the evaluation of healthcare resources. Studies have explicitly pointed out that focusing solely on the “quantity” of resources while neglecting their “structural heterogeneity” and ‘quality’ severely limits a comprehensive understanding of healthcare system effectiveness [41]. Chinese Health Service Management. However, this research does not merely adopt the cutting-edge approach of “constructing composite indicators” [42]; it further deepens its contextualization through significant contextualization. This enables assessments to not only reveal inequalities in resource distribution but also precisely identify the specific service types implicated in such inequalities. Consequently, it achieves a substantive advancement from measuring “resource abundance” to diagnosing “service gaps.”

At the level of demand assessment, this study employs a dynamic model that incorporates multiple spatiotemporal scenarios, which serves as a response to the limitations of classical static population data models. Static models are typically derived from census data or population distribution data from a single point in time. Such data cannot capture the spatiotemporal variation characterized by “centripetal agglomeration during the day and centrifugal dispersal at night” in residents’ activities. This study reveals significant differences in accessibility across various life scenarios, further corroborating the findings of Mao, L. [43], which suggest that neglecting demand dynamics may lead to planning inaccuracies. Although the use of big data—such as real-time travel duration [44] and Baidu data Bao, W. et al. [26,45] to capture spatiotemporal population dynamics has become a cutting-edge trend, most applications still focus on depicting “the trajectories of crowd movement.” The salient contribution of this research lies in its innovative integration of dynamic data streams with “typical life scenarios” that carry clear public policy implications. This integration allows the analysis to go beyond merely describing temporal changes in population distribution, enabling instead an interpretation of the spatiotemporal fluctuations and risks in potential healthcare demand under different daily behavioral patterns, such as commuting and leisure. For example, the finding that “accessibility is optimal during the morning commuting scenario” not merely describes a phenomenon but also reveals deeper equity issues: the current facility layout is closely coupled with urban employment centers, which may result in a spatial mismatch with the daily activity patterns of certain groups, such as the elderly population.

In calculating walking accessibility, this study employs real time walking duration data from the Gaode Map API combined with a Gaussian decay function, aiming to precisely simulate the spatial impedance of residents’ actual healthcare seeking behavior. This methodological choice aligns with the current forefront trend in healthcare accessibility research toward finer grained and more realistic modeling. Recent studies widely agree that moving beyond simple Euclidean distance to adopt complex cost measurements based on actual road networks is crucial for accurate assessment [46]. For instance, Zeng et al., when evaluating the spatial equity of multiple types of medical service facilities, also applied the Gaussian two step floating catchment area (G2SFCA) method and integrated facility type diversity—a methodological framework consistent with the core approach of this study [19]. A nationwide study further generated a hospital accessibility map at 1 km resolution, explicitly stating that refined travel time data form the basis of assessment [20]. This study’s adoption of real time walking duration is a direct response to such methodological calls.

The core empirical finding of this research is that the supply and demand of PHCI exhibit a “core–periphery” differentiation, with calculated Gini coefficients generally exceeding 0.5 (and reaching as high as 0.7 in some cases). This strongly “corroborates” the conclusions of several recent studies. First, the shortage of services in peripheral areas due to the concentration of medical resources in urban centers is a phenomenon repeatedly observed across multiple urban contexts. A micro-scale dynamic study on healthcare provision in Germany found that although overall accessibility improved over the long term, inter-regional disparities persisted [47], similarly revealing the stubborn nature of uneven spatial allocation—a result highly consistent with the findings here. Second, regarding the severity of inequity, the high Gini coefficients obtained in this study resonate with the latest survey focusing on remote rural areas in China, which points to significant and persistent inequalities in the allocation of doctors and equipment in primary healthcare [48]. Broader evidence comes from a scoping review of middle- and high-income countries, noting that residents in urban periphery areas generally have lower utilization rates of primary care services, often directly linked to inadequate accessibility [49]. Therefore, the present findings not only confirm the existence of severe spatial inequity in Shenyang but also, from the perspective of a Chinese megacity, add new and robust evidence to the global understanding that “spatial inequality in healthcare resources is a widespread challenge.”

The primary contribution of this study, however, lies in extending existing research paradigms. Many classic or contemporary accessibility studies either treat medical facilities as homogeneous supply points or rely on static population distribution data. For example, Wan et al. while assessing medical facility allocation within Chengdu’s “15 min life circles” and revealing an accessibility gap between the city center and the outskirts, based their analysis mainly on the overall layout of hospitals and multimodal transportation, without deconstructing the service capacity within institutions [50]. In contrast, the innovative breakthrough of this study is reflected in a coupled analysis along two dimensions:

Structural deconstruction on the supply side: This study does not treat PHCIs as monolithic entities but instead distinguishes the supply capacities of different basic medical services within them. This directly addresses the warning raised by Zeng et al. that overlooking the diversity of facility types can lead to an underestimation of both accessibility and equity [19]. The analysis confirms that even in areas with generally acceptable overall accessibility, access to specific services (e.g., pediatrics or chronic disease management) can still be a weak link. This reveals potential structural inequalities in resource allocation.

Spatiotemporal dynamization on the demand side: This study innovatively integrates dynamic population heat-map data with four typical life scenarios—“sleeping, morning work, afternoon work, and night leisure.” This elevates demand analysis from the static level of “population distribution” to the dynamic level of “spatiotemporal behavior.” Consequently, the research identifies a refined pattern of “supply–demand matching improved relatively during the morning work period”—a nuance completely beyond the reach of traditional static analyses.

The issue of insufficient walking accessibility to PHCIs revealed in this study stems from a systemic disconnect between planning/configuration methods and the dynamic, heterogeneous reality of urban environments. Building on this analysis, the following optimization strategies are proposed.

4.1. Shifting from an “Administrative Unit” to a “Population-Spatial Efficacy” Allocation Model

The configuration of PHCIs within the study area adheres to the Regulations on the Administration of Urban Community Health Service Institutions (No. 239, 2006), which bases PHCI establishment on subdistrict administrative boundaries. This approach considers primarily population size factors, neglecting the influence of spatial elements such as residential area scale and morphology on PHCI configuration. As illustrated in Figure 10, Community A and Community B have comparable total populations but significantly different areas. The walking accessibility to community health service stations is markedly weaker in Community A than in Community B. Community A’s irregular shape results in poor walking accessibility for residents in its peripheral areas trying to reach Community Health Station A. In response to this issue, recent research suggests that delineating “dynamic service areas” based on fine-grained spatiotemporal data can reflect actual needs more accurately than relying on fixed administrative boundaries [43]. Therefore, we recommend the following: First, incorporate “multi-scenario walking accessibility” assessment into local PHCI layout planning and annual evaluation systems, using it as a core basis for adjusting resource investment. Second, optimization should not rely solely on constructing new physical facilities. Within China’s predominantly public system, a more feasible pathway is to promote the flexible of resources. For example, a two-tier “central PHCI—service station” grid could be established, where core institutions provide personnel rotation, remote diagnostic support, and unified medication distribution to stations in low-accessibility communities. This enhances service capacity in peripheral communities while avoiding redundant construction and resource underutilization. This model of achieving “systematic efficiency gains” within the existing public network is more operational and sustainable, and better aligns with the healthcare reform direction of “strengthening primary care.”

4.2. Promoting Standardization and Complementarity in Service Provision

The current spatial layout of primary healthcare institutions fails to adequately account for how the uneven provision of different medical services affects accessibility. As shown in Figure 11, Communities A, B, and C enjoy comprehensive primary healthcare services, with good accessibility to all four service types. In contrast, Community D only provides Service 2, forcing its residents to seek other services at primary healthcare institutions in neighboring communities, resulting in significantly poorer accessibility. This study reveals that among 15 standard medical services, only 5 primary healthcare institutions can provide service A1, merely 2 can provide A2, 5 can provide B4, and 4 can provide B5. Furthermore, only 3 institutions are capable of providing both services C1 and C2 simultaneously. Consequently, residents may live in communities with a primary healthcare institution yet lack access to comprehensive care, often having to seek services across different areas, which compromises walking accessibility. This reflects a “structural problem” on the supply side. Heterogeneity and uncertainty in service content can severely undermine the overall efficiency and equity of the primary healthcare system [51]. Therefore, we recommend that municipal health authorities take the lead in formulating and issuing Community Basic Medical Service Standards aligned with the “15 min life circle” concept. These standards should clearly specify the mandatory and recommended service portfolios (e.g., general practice, pediatrics, chronic disease management, vaccination) for different tiers of PHCIs. Building on this, a hybrid “centralization and decentralization” strategy can be adopted [17]. Neighboring PHCIs should be encouraged to develop specialized services while ensuring the homogenization of basic services. Through appointment and referral mechanisms, services can be shared across communities, thereby fostering a regional service network that is both standardized and diversified.

4.3. Applying Digital Technologies to Respond to Spatiotemporal Demand Fluctuations

The current PHCI configuration, based on static population data, is entirely inadequate for addressing the tidal fluctuations in demand caused by job-housing separation and daily activity rhythms. From a spatial perspective, this leads to a dilemma rooted in uneven population distribution: two communities operate independently without considering the overall layout of the city’s public healthcare facilities, resulting in excessive concentration of resources. This creates a spatial mismatch—some areas suffer from a surplus of medical resources while others face shortages (Figure 12a). Temporally, healthcare demand varies across different scenarios such as sleep, work, and leisure, while the relatively fixed medical supply cannot adapt to these changing needs (Figure 12b). The L-H and H-L regions in Figure 13 more clearly illustrate the spatiotemporal characteristics of the mismatch between medical supply and demand. Recent methodological research on “time-varying accessibility” emphasizes that capturing the spatiotemporal dynamics of demand is crucial for resource optimization [18,52]. To address this, we propose the following: First, promote data sharing between urban planning and health departments to establish a “Community Health Demand Spatiotemporal Monitoring Platform.” This platform should integrate multi-source data, including land use, mobile phone signaling, and heat maps, enabling a shift from static planning to dynamic simulation. Second, in areas with high demand volatility (e.g., large residential complexes, industrial parks), promote “flexible service” models. These could include adjusting outpatient service hours based on commuting patterns or installing 24 h self-service pharmacies and smart diagnosis kiosks [53]. Third, Research has confirmed that telemedicine is a key technology for transforming healthcare delivery and improving accessibility [54,55,56]. The research area may adopt ‘Internet + Healthcare’ as its core strategy to achieve spatial-temporal equilibrium between supply and demand. Vigorously develop online follow-up consultations, medication delivery, and health consultations covered by medical insurance. This can effectively reduce residents’ absolute reliance on physical accessibility and serves as the most cost-efficient complementary strategy for addressing dynamic demand patterns.

Although the findings of this study originate from Shenyang, a classic monocentric and aging industrial city, the three core contradictions it identifies—the mismatch between administratively driven resource allocation and actual spatial demand, structural inequities in service provision, and the inability of static planning to accommodate dynamic urban rhythms—are common to many regions across China and globally. Therefore, the strategic principles proposed in this study, such as replacing static indicator evaluation with dynamic accessibility assessment, promoting standardization and complementarity in service delivery, and using digital technologies to bridge spatiotemporal gaps between supply and demand, possess broad transferability. When applying this approach, other cities can adapt the 2SFCA-MSD framework developed here by integrating it with local data to generate their own healthcare accessibility profiles. This will support the design of more targeted and scientifically informed plans for primary healthcare facility allocation, which constitutes the core practical value of this research methodology.

5. Conclusions

This study developed a spatiotemporal assessment framework (2SFCA-MSD) integrating multi-dimensional supply and dynamic demand to evaluate the walking accessibility and equity of Primary Healthcare Institutions (PHCIs). The main findings are as follows:

(1) Significant spatiotemporal mismatch exists between PHCI service supply and resident demand. Supply is excessively concentrated in old urban areas, leaving new urban zones underserved, while specific medical services (e.g., A2, C1) are severely lacking. Concurrently, demand exhibits substantial intra-day fluctuations, peaking around transportation hubs and commercial centers, misaligning with the static distribution of facilities. (2) Walking accessibility to PHCIs is severely inadequate and spatially inequitable. Only 45.7% of residential blocks in the study area are within a 15 min walking radius of a PHCI, with over 80% of the area classified as having the lowest accessibility. High-accessibility zones are confined to the urban core, creating “service-deficient areas” on the periphery. (3) This resource mismatch has led to severe spatial inequity. The Gini coefficients for all service types and living scenarios exceed 0.5, reaching up to 0.71, confirming a pronounced crisis in the equitable allocation of healthcare resources.

This study moves beyond traditional static analysis to successfully characterize the dynamic interplay between multi-type supply and time-sensitive demand. The core policy implication of this finding is the urgent need to reform current planning standards based on static administrative divisions and population data, shifting towards a new paradigm centered on dynamic accessibility. We argue that quantitative metrics, such as the “15 min walking catchment area,” should serve as the fundamental basis for PHCI siting and resource allocation. Consequently, this study not only reveals the complexity of actual healthcare access but also provides scientific evidence and an actionable implementation blueprint for achieving equitable and precise health spatial planning within the framework of the “15 min city”.

6. Research Limits and Prospects

This study has the following limitations: (1) This study utilizes Baidu Heat Map data, which effectively reflects the activity intensity of users within the Baidu ecosystem but does not constitute comprehensive population statistics. Although smartphone usage among the elderly population is becoming increasingly common, their preference for or frequency of using Baidu applications may be relatively limited. As a result, the data may still systematically underestimate the scale of their daily activities. Given that the elderly are a core user group of primary healthcare institutions, this potential bias could affect the accurate characterization of the spatial distribution of service demand. Future research could employ data sources with broader coverage and the ability to distinguish user attributes, such as age groups—for example, mobile signaling data—to more accurately assess actual disparities in access to healthcare resources across different population groups. (2) This study did not sufficiently account for residents’ willingness to seek medical care when calculating the accessibility of primary healthcare institutions. In actual healthcare-seeking behavior, residents’ choices are influenced not only by geographic distance but also significantly by the quality of medical staff—often opting to travel farther for a provider recognized for expertise in a specific condition. Subsequent research will require additional data to support a more comprehensive analysis of primary healthcare accessibility.