MilieuxVie: An Open-Source Web Mapping Tool for Assessing Context-Relative Service and Mobility Proximity for Complete-Neighbourhood Planning in Rural and Peri-Urban Municipalities

Abstract

Complete neighbourhoods, places where residents can meet their daily needs on foot, have become a central component of healthy and sustainable urban planning. Yet most assessment frameworks are calibrated for dense metropolitan environments, leaving rural and peri-urban municipalities without operational tools suited to their territorial needs. This article presents MilieuxVie, an open-source, browser-based interactive mapping application developed for the Laurentides health region of Québec (76 municipalities, 11 land-based unorganised territories, 2 indigenous territories and 4 aquatic administrative units; 93 territorial units in total; ~680,000 inhabitants). The tool evaluates the spatial accessibility of 12 service categories drawn from the Vivre en Ville (2026) complete-neighbourhood framework and OpenStreetMap data, using 2026 residential parcels from the provincial property assessment roll as origin points and weighting results by number of dwelling units. Three adaptive radius tiers (dense, intermediate, rural), based on residential dwelling-unit density (dwellings per km² of residentially designated urban land), scale the distance standards to settlement density. Because thresholds are scaled to settlement density, scores express context-relative service proximity rather than a uniform pedestrian standard and should not be read as directly comparable absolute accessibility across rural, peri-urban, and urban settings. A dedicated urban perimeter mode further disaggregates analysis to sub-municipal built-up zones, aligning the tool with Québec’s provincial Government land-use planning guidelines (GLPG). Gap analysis outputs identify which service types fall below the 70% coverage target, helping elected officials and planners identify where to focus further analysis. Results illustrate the scope of accessibility deficits across the region and highlight the analytical limits of uniform distance thresholds when applied beyond metropolitan contexts. Scores differ significantly across different settings (Kruskal–Wallis p = 0.006); the adaptive radius tiers narrow but do not close the structural gap, with rural municipalities scoring significantly lower than dense ones. The tool is freely available and requires no software installation, making it directly deployable by local planning offices.

1. Introduction

The spatial organisation of daily life—whether residents can reach food, healthcare, schools, parks and transit by walking—has become a central concern in public health, urban planning and climate policy [1,2]. Under the umbrella of the “complete neighbourhood” concept [3], a growing body of research links service proximity to physical activity levels [4,5,6], reduced automobile dependence [7], social cohesion [6,8,9] and health equity [10]. At the same time, the gain in popularity of the 15-min city concept [11] has renewed political interest in proximity-based planning and catalysed the adoption of measurable neighbourhood targets in cities around the world.

Most of the available methodological frameworks for assessing complete neighbourhoods, however, were developed in, and calibrated for, dense metropolitan environments [12,13,14]. When applied to rural or peri-urban contexts, standard walking-distance thresholds—typically 400 to 800 m for daily services—systematically underestimate actual accessibility, producing uniformly poor scores that fail to differentiate between territories and offer little guidance for local planning decisions [15,16,17,18,19]. This methodological gap is particularly consequential in Québec, where the Government land-use planning guidelines (GLPG) [20] frame provincial land-use policy around the consolidation of urban perimeters in municipalities of all sizes, yet operational tools for measuring the completeness of those perimeters remain scarce outside major urban centres.

Several open-source tools exist for related purposes. The Walk Score API [21] provides point-level walkability estimates but does not disclose its methodology, requires payment for bulk use, and does not distinguish rural contexts. Existing isochrone/walkshed tools in research routinely intersect reachability polygons with points (services) and lines or grids (population), then aggregate these to various spatial units. Several workflows already summarize results by districts, towns, or hexagonal tiles, and tool architectures (e.g., Integrated Services Optical Grid Architecture (ISOGA) [22,23], CityChrone [24,25,26], catchment and travel analysis (CTA) [27,28] are explicitly built to join isochrones with external socio-spatial datasets. While out-of-the-box APIs like OpenRouteService or Valhalla focus on geometry, the literature shows clear, replicable patterns for aggregating service coverage up to municipal or sub-municipal units using Geographic Information Systems (GIS) operations on top of those isochrones [29]. Tools specifically designed for Québec’s institutional context, cross-referencing provincial property assessment data with planning unit boundaries, are largely absent from the literature.

This paper describes MilieuxVie, an interactive web mapping application developed at the Public health department of the Laurentides Health Region as part of knowledge-transfer activities targeting elected municipal officials and land-use planners in the Laurentides health region. The tool operationalises the 15-parameter complete-neighbourhood framework of Vivre en Ville [3] using freely available spatial data (OpenStreetMap via the Overpass API) and the provincial residential property roll [30]; it offers adaptive radius thresholds calibrated to three territorial types, and generates interpretable gap analyses at both the municipal and urban-perimeter levels. The application is distributed as a single self-contained HTML file requiring no installation, no server infrastructure, and no proprietary data. The methodological contribution of this work is not only the software per se but the combination of dwelling-weighted residential-parcel origins, milieu-adaptive thresholds, and urban-perimeter disaggregation into a single reproducible framework calibrated for the rural–urban continuum; the browser-based tool is the means by which this method is made usable in planning offices without GIS capacity.

The remainder of the article is structured as follows. Section 2 reviews the conceptual and methodological background. Section 3 describes the study area, data and technical architecture. Section 4 presents the accessibility assessment methodology in detail. Section 5 illustrates the tool’s outputs for the Laurentides region. Section 6 discusses the results in the context of existing frameworks and outlines directions for future development. Section 7 concludes.

2. Background

2.1. Complete Neighbourhoods and the 15-Min City

The concept of a complete neighbourhood refers to a residential environment in which residents can satisfy their fundamental daily needs within a short distance of their home, regardless of their income or mobility status [3,31,32]. While precursors can be found in early-twentieth-century new urbanism and the Transit-Oriented Development literature [33], the concept gained mainstream policy visibility through Moreno et al.’s formalisation of the 15-min city [11,34,35], subsequently adopted as a planning objective in Paris, Melbourne, and a growing number of C40 member cities.

In the North American context, Boisjoly et al. [35] proposed a “30-min city” adaptation acknowledging lower residential densities and greater transit travel times, arguing that polycentric metropolitan structures require a larger spatial envelope. Vivre en Ville [3] translated this concept into an operational 15-parameter framework for Québec municipalities, distinguishing three categories of attributes: housing diversity, proximity destinations, and mobility infrastructure. For each destination type, the framework specifies a target walking radius calibrated to an assumed urban density of 25–30 households per hectare.

2.2. Accessibility Measurement: Methods and Limitations

Spatial accessibility to services can be measured along a spectrum from simple straight-line (Euclidean) buffers to network-based travel-time isochrones and gravity-model catchment areas [36]. While network-based approaches are methodologically superior, their computational demands and reliance on complete street-network data limit their applicability at the scale of a health region with 93 territorial units (76 municipalities and 17 non-municipal entities), several of which lack complete OpenStreetMap (OSM) road coverage [37]. Moreover, the difference between Euclidean and network distances tends to diminish in grid-like street layouts common in Québec’s small urban centres [38].

A fundamental methodological choice concerns the origin point of accessibility calculations. Many studies use evenly spaced population grids or census dissemination area centroids, which may misrepresent the actual spatial distribution of residents—particularly in municipalities with concentrated settlement patterns surrounded by large agricultural or forested areas [35]. Using geocoded residential parcels as origins, weighted by dwelling unit count, provides a more accurate representation of the population actually exposed to (or deprived of) service proximity.

The applicability of urban accessibility thresholds to rural settings has received increasing attention. Apparicio et al. [39] documented systematic under-performance of proximity indicators in low-density suburbs. Authors such as Lister [40] have argued for context-sensitive threshold adaptation as a prerequisite for meaningful rural–urban comparison. No consensus has emerged on the appropriate scaling factors, however, and most operational tools continue to apply uniform thresholds. To address this limitation, the framework developed in this study explicitly adopts a context-relative approach. Rather than measuring absolute pedestrian accessibility under a uniform walking standard, it assesses service proximity against expectations scaled to local settlement density. Consequently, the resulting scores express relative fulfillment of proximity needs and are not designed for direct absolute comparison across varying rural, intermediate, and urban settings.

2.3. Volunteered Geographic Information and OpenStreetMap in Planning Contexts

OpenStreetMap (OSM) has become a primary data source for service-proximity research, particularly where authoritative point-of-interest datasets are unavailable or commercially restricted [41]. Studies have generally found that OSM completeness is sufficient for most amenity categories in urban areas but declines substantially in rural and peri-urban municipalities [42,43]. This unevenness constitutes a significant limitation for cross-territorial comparisons, as a low accessibility score may reflect genuine service deficits or simply incomplete OSM coverage.

The Overpass API provides programmatic query access to the full OSM database, enabling targeted extraction of service features within user-defined geographic areas [44]. Recent work has demonstrated the feasibility of regional-scale Overpass queries encompassing hundreds of square kilometres in sub-min query times, making real-time browser-based accessibility calculation technically viable [45]. Overpass is widely used to extract building footprints, coordinates, and road networks for urban analytics and planning workflows, which implies the retrieved geometry is often operationally useful even when richer attributes are missing [46,47,48].

3. Study Area, Data and Technical Architecture

3.1. Study Area

The study area is the health region of the Laurentides (Québec, Canada), comprising 93 local territorial units organised into 8 regional county municipalities (MRCs) and covering approximately 21,559 km². Following the Ministry of the Environment, Fight Against Climate Change, Wildlife and Parks territorial classification, the 93 units comprise 76 municipalities, 15 unorganised territories (which 4 are purely aquatic administrative units with no land area, and 11 are land-based unorganized territories), and 2 indigenous territories (Kanesatake Mohawk Territory and Doncaster Reserve). The population of approximately 680,000 is distributed very unevenly: the southern tier (MRCs of Deux-Montagnes and Thérèse-De Blainville) contains dense peri-urban municipalities of the greater Montréal metropolitan area, while the northern tier (Antoine-Labelle) contains municipalities exceeding 1000 km² with settlement clusters separated by forests and lakes. This gradient of density and urbanisation makes the region an appropriate test case for a methodology that must perform well across the full rural–urban continuum.

3.2. Data Sources

Four primary data sources were integrated (

Table 1. Data sources integrated in MilieuxVie.

Source	Data Type	Year	Use in MilieuxVie
Ministry of Municipal Affairs and Housing property roll	Residential parcels (pts, n = 234,466; 305,827 units)	2026	Origin points; dwelling-unit weighting
OpenStreetMap/Overpass API	Points of interest (12 service categories)	Real-time	Destination locations
Ministry of the Environment, Fight Against Climate Change, Wildlife and Parks	Local territorial unit boundaries and areas (93 polygons: 76 municipalities, 15 unorganized territories, 2 indigenous territories)	2024	Spatial units; milieu-type classification
Ministry of Municipal Affairs and Housing	Urban perimeter polygons (92 features, 53 munis)	2024	Sub-municipal analysis unit
Statistics Canada	Census housing typology (tenure, bedroom distribution)	2021	Housing dimension (4 parameters)
Quebec Housing Corporation	Subsidised housing provider locations (41 organisations)	2026	Affordability cartographic layer

). The provincial residential property assessment roll [30] provides the location and unit count of all residential buildings (Property Code 1000–1590) in the province. For the study region, this yielded 234,466 residential parcels representing 305,827 dwelling units. OpenStreetMap (OSM) data were retrieved via the Overpass API using a bounding-box query covering the entire region (~258 × 180 km²) at analysis time. Municipal boundaries and areas were obtained from the Ministry of the Environment, Fight Against Climate Change, Wildlife and Parks. Urban perimeter polygons were supplied by the regional planning authority under the Ministry of Municipal Affairs and Housing. Supplementary data on housing typology (bedroom distribution, tenure proportions) were derived from the 2021 Statistics Canada census, and an inventory of subsidised housing providers was obtained from the Quebec Housing Corporation. All spatial layers are stored and processed in a common geographic coordinate system (WGS 84, EPSG:4326); no on-the-fly reprojection is performed. Proximity distances are computed directly on geographic coordinates using the haversine formula, and the web map renders the layers in Web Mercator (EPSG:3857) for display purposes only.

3.3. Technical Architecture

MilieuxVie is implemented as a single self-contained HTML file (~11 MB). No web server, database, or external authentication is required: the file can be opened directly in any modern browser. The file runs from the local filesystem (file://) without a web server; internet access is required for the mapping library and basemap tiles (CDN) and for live Overpass queries. Importing a saved GeoJSON (e.g. QGIS 4.0 [49]) uses the browser file picker and needs no server. In benchmarking (n = 30 consecutive regional runs), the end-to-end computation completed in a median of 160 s (interquartile range: 142–295 s; range: 115–575 s; no failures). Tests were conducted on a Windows machine (13th Gen Intel Core i5-1335U 1.30 GHz, 10-core, 16 GB RAM) using Google Chrome v.149 over a 4G network (10 Mbps) (Supplementary Table S1). The live Overpass query accounted for a median of 147 s (measured client-side from request dispatch to complete receipt of the JSON response, and therefore including server queue wait, query execution, network transfer of the multi-megabyte response, and browser-side parsing; query execution itself remains within the 180-s server timeout of the regional query), about 92% of the total, while the tool’s local computation was a small and stable fraction (median ≈ 16 s, falling to ≈ 3 s on a warm cache). End-to-end time is therefore dominated by, and varies with, Overpass server load rather than the client device.

This deployment model was chosen to maximize accessibility within municipal planning offices, which frequently operate under restrictive IT policies that preclude the installation of GIS software (e.g., QGIS 4.0 [49]) or the use of externally hosted platforms.

The front-end stack consists of Leaflet.js 1.9.4 for cartographic rendering (CARTO Positron basemap), vanilla JavaScript for computation and User Interface (UI), and the Overpass API for on-demand OpenStreetMap (OSM) queries. Pre-computed datasets (residential parcel centroids, urban perimeters, housing typology, affordable housing providers) are embedded as compressed JSON objects within the file itself, eliminating external data dependencies for the residential origin and housing dimensions. OSM service-facility data are retrieved at analysis time via Overpass, enabling queries to reflect the current state of the OSM database.

A key optimisation is the use of a single regional bounding-box query (covering the full Laurentides extent, ~258 × 180 km² rather than 93 individual per-local territory queries. The regional query uses a 180-s server-side execution timeout (the Overpass [timeout:] parameter, which bounds query execution on the server only), per-municipality by-click queries use a 90-s timeout, and all requests fall back automatically across three public Overpass endpoints on failure. Subsequent per-municipality or per-urban-perimeter calculations are performed entirely in the browser by filtering the cached feature set to the relevant polygon using a ray-casting point-in-polygon algorithm.

4. Methodology

4.1. Service Categories and Distance Thresholds

Table 2. Indicator hierarchy and reconciliation of the Vivre en Ville framework with MilieuxVie.

Macro-Dimension	Vivre en Ville Reference Dimension	MilieuxVie Parameter (SMP)	Operationalisation—OpenStreetMap Tags	Status
Housing (Habitation)	Housing (four parameters): Size diversity (diversité de taille) Rental share (part locative) Availability (disponibilité) Affordability (abordabilité)	Housing block (four params)	Provincial assessment roll, ISQ household projections, and census data (not OSM).	Excluded from service- and mobility-proximity (SMP) composite—reported separately pending validation
Proximity Destinations (Destinations de Proximité)	Food (alimentation)	Food retail	shop = supermarket, convenience, grocery, butcher, bakery, greengrocer, deli, food, general	Retained
	Childcare services (service de garde)	Childcare	amenity = kindergarten, childcare	Retained
	Schools (établissements scolaires)	Elementary school	amenity = school (primary; by name)	Split
	Schools (établissements scolaires)	Secondary school	amenity = school (secondary)/college/university	Split
	Nature (nature)	Green/natural space	leisure = park, garden, nature_reserve; land use = forest, grass, meadow	Retained
	Leisure and socialisation (loisirs et socialisation)	Recreation and sport	leisure = sports_centre, fitness_centre, swimming_pool, stadium, pitch, recreation_ground, ice_rink	Merged
	Complementary activities (activités complémentaires)	Cultural facilities	amenity = theatre, cinema, library, community_centre, arts_centre, social_centre	Merged
	Healthcare (soins de santé)	Pharmacy	amenity = pharmacy	Split
	Healthcare (soins de santé)	Primary healthcare	amenity = clinic, hospital, doctors, health_centre, dentist	Split
Mobility (Mobilité)	Pedestrian infra. (infrastructures piétonnières)	N/A	Operationalised implicitly via distance calculation algorithms rather than as a destination feature.	Excluded
	Cycling infra. (infrastructures cyclables)	Cycling network	highway = cycleway; bicycle = designated/yes	Retained
	Public transit (transport en commun)	Public transit	highway = bus_stop; amenity = bus_station, ferry_terminal; railway = station, halt, tram_stop	Merged
	Shared mobility (mobilité partagée)	Shared mobility	amenity = car_sharing, bicycle_rental	Retained

Notes: The Vivre en Ville complete-neighbourhood framework comprises 15 reference parameters grouped into three macro-dimensions. MilieuxVie translates these into 16 distinct indicators: 12 mappable OSM service categories and a four-dimension housing block. Specifically, the Vivre en Ville Schools and Healthcare dimensions were split to reflect their distinct spatial distributions in peri-urban and rural contexts, while Pedestrian infrastructure was excluded as a mapped parameter because walkability is intrinsically measured by the tool’s spatial distance calculations. The Housing block is currently excluded from the service- and mobility-proximity (SMP) composite pending cross-regional data validation, which is why the composite is named SMP score rather than a full complete-neighbourhood score.

presents the conceptual mapping that combines the Vivre en Ville (2026) [3] complete-neighbourhood framework with the MilieuxVie operationalisation. The Vivre en Ville framework is structured around three macro-dimensions: Housing (Habitation), Proximity destinations (Destinations de proximité), and Mobility (Mobilité). The 12 service and mobility categories enter the service- and mobility-proximity (SMP) composite score. The four housing-dimension parameters (dwelling-size diversity, rental proportion, vacancy rate, long-term affordability) require census and Canada Mortgage and Housing Corporation (CMHC) data and are handled separately; the tool currently reports these parameters descriptively rather than integrating them into the composite score.

Each OSM category is matched to one or more tag combinations using a rule-based classifier applied to element tags returned by the Overpass query (

Table 3. OSM service categories, and adaptive distance thresholds by milieu type.

Service Category	Dense (≥100 Dwellings/km2)	Intermediate (≥10–<100 Dwellings/km2)	Rural (<10 Dwellings/km2)
Food retail	800 m	1500 m	3000 m
Childcare	400 m	800 m	1500 m
Primary school	800 m	1500 m	3000 m
Secondary school	1600 m	3000 m	5000 m
Green/natural space	400 m	800 m	1500 m
Recreation and sport	800 m	1500 m	3000 m
Cultural facilities	800 m	1500 m	3000 m
Pharmacy	400 m	800 m	2000 m
Primary healthcare	800 m	1500 m	5000 m
Public transit	800 m	1500 m	3000 m
Shared mobility	800 m	1500 m	3000 m
Cycling network	400 m	800 m	1500 m

). In alignment with the context-relative approach discussed previously, these adaptive distance thresholds scale with local settlement density rather than enforcing a uniform pedestrian standard. A 400 m threshold in a dense setting and a 1500 m threshold in a rural setting both represent a target level of proximity given their respective territorial realities, though they do not offer the same absolute walkability. Each OSM category is matched to one or more tag combinations using a rule-based classifier applied to element tags returned by the Overpass query (Table 3).

4.2. Milieu-Type Classification

Three milieu types are distinguished based on residential dwelling-unit density (dwellings/km²), computed by dividing the number of dwelling units from the Ministry of Municipal Affairs and Housing 2026 property roll by the area placed under residential urban land-use designation in the regional county municipality (MRC) land-use and development plans. Restricting the denominator to the residentially designated area, rather than the total municipal area, provides a more direct proxy for settlement intensity and avoids the dilution caused by the large uninhabited territories common in the Laurentides region. Dense municipalities (≥100 dwellings/km², n = 29) receive the standard Vivre en Ville thresholds. Intermediate municipalities (≥10–<100 dwellings/ km², n = 32) receive thresholds scaled approximately 1.5–2×. Rural municipalities (<10 dwellings/km², n = 17) receive thresholds scaled approximately 3–6×. These scaling factors were informed by the rural accessibility literature [36,50,51] and by informal, unstructured discussions with regional planners (not a structured elicitation); their principal justification is the sensitivity analysis reported in Section 6. The density-based approach correctly reclassifies municipalities such as Saint-Jérôme (444 dwellings/km², dense) and Mont-Laurier (37 dwellings km², intermediate) that were misclassified under an area-only criterion. Sensitivity to the choice of thresholds is discussed in Section 6.

4.3. Residential-Parcel Accessibility Score

For each municipality and each service category, the parameter score is computed as the proportion of dwelling units whose nearest facility of that type lies within the threshold distance:

Scorep = ∑ u_i I(d_i,p ≤ r_p]/∑ u_i

(1)

where u_i is the number of dwelling units in parcel i, d_i,p is the Haversine distance from parcel i to the nearest OSM feature of type p, and r_p is the adaptive threshold for parameter p given the municipality’s milieu type. The composite service- and mobility-proximity (SMP) score is calculated as the arithmetic mean of all 12 parameter scores. Equal weighting is adopted as a deliberately transparent, value-neutral default in the absence of an empirically validated weighting consensus; it is a simplification whose influence is examined in the sensitivity analysis (Section 6.2). Because the arithmetic mean is fully compensatory, we also report a non-compensatory companion: the number of the 12 dimensions meeting the 70% target, together with a minimum-performance flag raised when any essential service (food retail, primary healthcare, childcare, pharmacy or schools) falls below target. Municipalities are classified as complete/promising (score ≥ 70%), developing (40–69%) or incomplete (<40%).

4.4. Urban Perimeter Analysis

Restricting the origin set to parcels within the urban perimeter raises scores because it alters the residential composition entering Formula (1), rather than removing uninhabited land from a denominator. When using parcel-weighted origins, uninhabited territory affects only the dwelling-density value used for setting classification. For municipalities with delineated urban perimeters (n = 53, 92 Urban Perimeters (UP) polygons), a sub-municipal analysis mode restricts the origin set to parcels within the UP boundary. This addresses an origin-set composition effect: a municipality with one concentrated village core surrounded by a large, forested area will receive a low composite score if all parcels are included, even if the village core itself has good service proximity. The UP-level score uses the same formula (Formula (1)) but with the summation restricted to parcels i ∈ UP. Service features are retrieved within a buffer equal to the maximum parameter threshold around the UP bounding box to ensure that facilities immediately outside the polygon boundary are not excluded. Both at the municipal and perimeter scale, services in neighbouring municipalities are included whenever they lie within the adaptive radius of a residential parcel, since the regional Overpass query retrieves all features within the full Laurentides bounding box irrespective of administrative boundaries. The accessibility calculation is parcel-centric, not perimeter-centric: for any origin parcel, every service within its adaptive radius is counted regardless of administrative boundaries. Services inside an urban perimeter are counted for parcels outside it when within range, and services outside a perimeter are counted for parcels inside it on the same basis. Perimeter delineation restricts the set of origin parcels considered, not the set of eligible destinations.

4.5. Gap Analysis

For each municipality or urban perimeter, the gap analysis identifies service categories with a parameter score below the 70% target. Parameters are ranked by the size of the gap (target − score) in descending order, providing a prioritised action list. These thresholds are interpretive conventions rather than empirical breakpoints; a unit scoring 69% is not categorically distinct from one scoring 70%, and the bands should be read as a continuum. For MRC-level reporting, the gap analysis aggregates across all computed urban perimeters within the MRC and reports the number and proportion of perimeters below the target for each service type, enabling regional health authorities to identify systemic territorial deficits that transcend individual municipalities.

4.6. Statistical Analysis

Non-parametric statistical tests were employed to validate the model’s parameters and analyse regional trends, as the score distributions did not meet normality assumptions. First, a Kruskal–Wallis test was used to compare the composite service- and mobility-proximity (SMP) scores across the three defined milieu types (dense, intermediate, and rural). Second, a Wilcoxon signed-rank test was conducted on the paired sample of municipalities and their respective urban perimeters to evaluate variations resulting from sub-municipal disaggregation. Third, Spearman’s rank correlation was applied to assess the geographic gradient between composite scores and centroid latitudes, as well as to validate the alignment between straight-line (Euclidean) distances and pedestrian-network distances derived via Open Source Routing Machine (OSRM). Statistical computations, including non-parametric tests and effect size estimations, were performed using Python and the SciPy statistical library (Supplementary File S2). Data processing and script generation were supported by Anthropic; models Claude Sonnet 5, 2026 [52,53].

5. Results

5.1. Tool Interface and User Modes

MilieuxVie is accessed as a single HTML file opened in a web browser. The interface is divided into two zones: a 400-pixel sidebar containing all controls, parameters and results panels, and an interactive Leaflet map occupying the remainder of the screen. The 93 territorial units of the Laurentides region are displayed on load as light green polygons against a CARTO Positron basemap. Four analysis modes are accessible from a persistent toolbar at the top of the sidebar (Figure 1).

The By-click mode (“On click”) launches a per-municipality analysis triggered by clicking a polygon on the map. An Overpass API query is sent for the selected municipality’s bounding polygon and the result is returned in 10–30 s depending on network conditions. Once the regional cache has been populated (see below), subsequent by-click analyses are instantaneous. The Regional mode (“Show all”) dispatches a single bounding-box query over the full Laurentides extent (~258 × 180 km²) and computes scores for all 93 territorial units in the browser without further server requests. Estimated completion time is 2–3 min, varying with Overpass server load and network conditions. The Load mode (“Charger”) imports a previously exported GeoJSON file and restores all scores on the map instantaneously, enabling offline use and presentation without requiring Overpass. The Urban Perimeter mode (“Perimeters”) activates a dedicated panel displaying the 92 urban perimeter polygons as a separate cartographic layer; municipal boundaries fade to near-transparent so that the perimeters become the primary visual unit. A single button triggers scoring for all 92 perimeters using the regional Overpass cache if already populated.

For each analysed unit—whether a municipality or an urban perimeter—the results panel reports: the composite SMP score (0–100%); a classification label (complete/promising ≥70%; developing 40–69%; incomplete <40%); the detected milieu type and adaptive radius tier; a comparative bar against the MRC median; individual scores for each of the 12 OSM parameters; a housing dimension summary drawn from census data; and a gap analysis listing all parameters below the 70% target in descending order of deficit, with contextual action guidance. Two export formats are available after a regional analysis: a GeoJSON file retaining full polygon geometry with one attribute per parameter (suitable for GIS import into QGIS 4.0 [49] or ArcGIS Pro 3.4 [54]), and a JSON tabular file suitable for Excel 16.0 [55].

5.2. Regional Overview: Municipal-Level Scores

The regional analysis was completed on 22 June 2026 and yielded SMP scores for 78 of the 93 territorial units (83.9%). The 15 units without a score are 4 aquatic unorganised territories with no residential land parcels, 9 land-based unorganized territories (named lake territories such as Lac-de-la-Pomme and Lac-Ernest), and Kanesatake and Doncaster Territories. Two land-based unorganized territories (Lac-Bazinet and Lac-Oscar) are scored at 0%. These units are displayed in grey on the map and excluded from all subsequent calculations.

Among the 78 scored territorial units, the composite SMP score ranged from 0 to 60%, with a median of 15% and a mean of 20.2% (SD = 16.5). No municipality reached the 70% completeness threshold. Fifteen territorial units (19.2%) were classified as “developing” (40–69%), and 63 (80.8%) as “incomplete” (<40%). The highest score was 60%, recorded for a municipality in the MRC of Thérèse-De Blainville. Kanesatake Mohawk Territory could not be reliably scored owing to substantial missing input data and an unreliable recorded area (which yields an undefined dwelling density); it is treated as a no-data unit, displayed in grey and excluded from the scored statistics.

Table 4. Distribution of composite SMP scores by milieu type, Laurentides region, 2026.

Milieu Type (Density Criterion)	n	Mean (%)	Median (%)	Min–Max (%)	Developing/Incomplete
Dense (≥100 dwellings/km2)	29	27.3	26.0	1–60	10/19
Intermediate (≥10–<100 dwellings/km2)	32	19.3	16.5	2–49	5/27
Rural (<10 dwellings/km2)	17	9.8	10.0	0–31	0/17
All scored municipalities (78)	78	20.2	15.0	0–60	15/63

presents the score distribution by milieu type.

Scores differed significantly across different settings (Kruskal–Wallis H = 10.29, df = 2, p = 0.006, ε² = 0.11): median scores rose from 10% in rural municipalities to 16.5% in intermediate and 26% in dense sectors. The adaptive radius tiers therefore narrow, but do not eliminate, the structural rural–urban gap, which remains statistically significant; the composite score also declined significantly from south to north (Spearman ρ = −0.36, p = 0.001, n = 78) (see Supplementary Table S3 for full test statistics). MRC-level medians revealed a marked north–south gradient (

Table 5. Median and mean composite SMP scores by MRC, Laurentides region, 2026.

MRC	n	Mean (%)	Median (%)	Range (%)
Mirabel	1	32.0	32.0	32–32
Thérèse-De Blainville	7	46.1	44.0	37–60
Deux-Montagnes	8	31.5	28.0	9–55
Les Pays-d’en-Haut	10	15.1	14.5	0–42
La Rivière-du-Nord	5	15.8	10.0	4–41
Les Laurentides	20	18.2	13.5	0–49
Antoine-Labelle	19	13.7	12.0	0–43
Argenteuil	9	13.6	5.0	1–47

): the two southernmost MRCs (Thérèse-De Blainville, Mirabel) recorded the highest median scores (44% and 32%, respectively), while the northern MRCs of La Rivière-du-Nord and Argenteuil recorded the lowest (10% and 5%).

5.3. Parameter-Level Patterns

At the parameter level, the analysis revealed a clear hierarchy of accessibility across the 12 service categories (

Table 6. Mean and median scores (% dwelling units within target radius) for 12 OSM service parameters across 78 scored territorial units (excluding Kanesatake; see text) and 89 scored urban perimeters, Laurentides 2026. Parameters ranked by municipal median score (descending).

Parameter	Muni Mean %	Muni Median %	Munis ≥ 70% (n/78)	PU Median %	PUs ≥ 70% (n/89)	Rank (Gap)
Food retail	43.0	46	13/78	90	61/89	12
Natural/green space	38.4	37	8/78	75	54/89	11
Primary school	30.7	29	8/78	53	41/89	10
Cycling network	31.9	26	16/78	0	20/89	9
Secondary school	10.3	0	4/78	0	7/89	4
Public transit	24.8	0	15/78	0	24/89	7
Cultural facilities	23.4	22	3/78	0	28/89	8
Healthcare (primary)	11.8	0	0/78	0	10/89	6
Recreation and sport	11.7	0	2/78	0	12/89	5
Pharmacy	8.5	0	0/78	0	4/89	3
Childcare	4.2	0	0/78	0	0/89	1
Shared mobility	2.9	0	1/78	0	2/89	2

). Food retail was the best-performing dimension at the municipal scale, with a mean score of 43.0% and 13 of 78 municipalities (16.7%) reaching the target. The cycling network ranked second (mean 31.9%), reflecting the relatively dense trail infrastructure in many Laurentides municipalities and the generous 400 m radius applied in dense zones. Primary school access achieved a mean score of 30.47% and natural/green spaces 38.4%.

At the opposite end of the spectrum, shared mobility (car-sharing and bicycle rental) returned the lowest mean score (2.9%), followed by childcare (4.2%) and pharmacy (8.5%). Notably, childcare and pharmacy scored zero in a majority of municipalities: 0 of 78 reached the 70% threshold. These results reflect two distinct phenomena. First, shared mobility infrastructure is genuinely sparse throughout the region. Second, while pharmacies and childcare facilities exist, they are typically isolated. They serve only their immediate surroundings, failing to cover a meaningful proportion of residential areas within short target radii (e.g., 400 m in dense settings). Public transit showed the highest inter-municipal variance (0–100%), reflecting the contrast between well-served peri-urban municipalities adjacent to other networks and rural municipalities with no scheduled transit service.

5.4. Urban Perimeter-Level Score

Of the 92 urban perimeters, 89 (96.7%) were successfully scored; three returned no residential parcels within their boundaries and were excluded. The 89 scored perimeters cover 161,544 residential dwelling units. Composite PU scores ranged from 0 to 79%, with a median of 26% and a mean of 28.8% (SD = 18.9). Two perimeters (2.2%) reached the 70% completeness threshold: Mont-Tremblant (Les Laurentides, 74%) and Sainte-Adèle (Les Pays-d’en-Haut, 74%). Although both perimeters score 74%, they reach it through opposite gap profiles: Sainte-Adèle’s cycling-network proximity is 39% against Mont-Tremblant’s 100%, while their food-retail proximity is comparable (90% and 100%). That two units share a score through different deficits illustrates a structural limitation of any compensatory index—the composite is best used as an entry point to the parameter-level gap profile, where the actionable planning information resides. Twenty perimeters (22.5%) were classified as developing (40–69%) and 67 (75.3%) as incomplete (<40%).

Compared to the municipal-level analysis, urban perimeter scores were systematically higher (median 26% vs. 15%; mean 28.8% vs. 20.2%), reflecting the restriction of the origin set to parcels within the perimeter—a change in residential composition rather than the removal of uninhabited land (with dwelling-weighted parcel origins, uninhabited territory cannot dilute the score). Because municipal and perimeter scores are computed on different spatial units, their comparison is subject to the modifiable areal unit problem (MAUP): the higher perimeter values partly reflect this origin-set restriction rather than genuinely better accessibility, and the two are not directly comparable. A paired test confirmed the difference (Wilcoxon signed-rank on the 53 municipalities with a perimeter: V = 101, p < 0.001, r = 0.72; the perimeter scored higher in 48 of 53 cases) (see Supplementary Table S3 for full test statistics). Both municipalities reaching the completeness threshold at the perimeter level had urban cores that were so dense and well-served that a majority of residential parcels within the urban boundary could reach most service categories within the adaptive target radii. Notably, Sainte-Adèle achieved a PU score of 74% while its municipal score was 42%, illustrating the analytical gain of the sub-municipal disaggregation. Computing each perimeter’s milieu from its own dwelling density, rather than inheriting the municipal class, reclassifies 47 of the 89 scored perimeters (53%), almost all toward a denser class; the perimeter scores reported here use this perimeter-scale classification.

At the parameter level across urban perimeters, food retail remained the best-performing category (median 96% across PUs), followed by natural/green space (76%) and primary school (62%). Childcare was the most universally deficient category, with all 89 scored perimeters falling below the 70% threshold (median 0%), followed by shared mobility (87/89 PUs below target) and pharmacy (85/89). These three dimensions constitute the most consistent regional gaps irrespective of municipality size or milieu type.

MRC-level gap summaries in the urban perimeter mode sidebar (Figure 2) reveal systematic territorial patterns: in the MRC of Antoine-Labelle, childcare, secondary schooling, and shared mobility were identified as priority gaps in all 22 scored perimeters. In the MRC of Thérèse-De Blainville, the patterns were more heterogeneous, reflecting the mix of dense suburban municipalities with good pharmacy and transit access alongside newer peri-urban developments with deficient cycling and shared mobility infrastructure.

5.5. Illustrative Case Studies: Two Urban Perimeters at the Completeness Threshold

To illustrate the analytical depth available at the urban perimeter scale, this section presents two municipalities whose urban cores achieved the 70% completeness threshold: Sainte-Adèle (MRC des Pays-d’en-Haut, périmètre #80, 74%) and Mont-Tremblant (MRC des Laurentides, périmètre #60, 74%). Both cases illustrate how the same composite score can arise from different gap profiles.

5.5.1. Sainte-Adèle—A Mixed-Use Mountain Town in Les Pays-d’en-Haut

Sainte-Adèle is classified as intermediate (77 dwellings/km², area 132 km²). Its main urban perimeter (#80) covers 2294 residential buildings and 3445 dwelling units. At the municipal scale, Sainte-Adèle scored 42%; at the perimeter scale, this rises to 74%, a 32-point improvement illustrating the origin-set restriction effect, whereby the perimeter score counts only the parcels in the built-up core (Figure 3).

Eight of the twelve parameters reached the 70% target: secondary school (98%), recreation and sport (93%), primary school (91%), food retail (90%), public transit (90%), cultural facilities (87%), natural/green spaces (80%), and primary healthcare (78%). This broad-spectrum provision reflects Sainte-Adèle’s role as a regional service centre in Les Pays-d’en-Haut.

Four parameters fell below the threshold. Shared mobility scored 34% (gap: 36 points), the cycling network 39% (gap: 31 points), pharmacy 53% (gap: 17 points), and childcare 56% (gap: 14 points). The childcare gap suggests that targeted facility siting could push this parameter across the threshold.

5.5.2. Mont-Tremblant—A Tourist Municipality with a Compact Urban Core

Mont-Tremblant (MRC des Laurentides) covers 130 km², intermediate (37 dwellings/km²). Its main urban core (#60, 2283 buildings, 3774 units) scored 74% (Figure 4), while four secondary perimeters scored 6–53%.

Seven parameters achieved the target: food retail (100%), secondary school (100%), public transit (100%), cycling network (100%), natural/green spaces (99%), recreation and sport (93%), and primary healthcare (92%).

Five parameters fell below the threshold. Shared mobility scored 0% (OSM under-mapping). Pharmacy scored 30% (gap: 40 points). Cultural facilities 53%, childcare 55%, primary school 62%. Childcare and school gaps may point to land-use levers such as zoning, though the appropriate response depends on non-spatial factors, demand, funding and licensing, not captured by the model. That two perimeters reach the same score through different deficits illustrates a structural limitation of any compensatory index: the score is best used as an entry point to the parameter-level gap profile, which is where actionable planning information resides.

Comparing both cases: shared mobility and childcare are common priority gaps. Pharmacy is more acute in Mont-Tremblant (30%) than Sainte-Adèle (53%). The cycling gap in Sainte-Adèle (39%) contrasts with Mont-Tremblant’s perfect score (100%), illustrating how the same 74% composite score can correspond to different underlying gap profiles, and hence different questions for planners to investigate.

6. Discussion

6.1. Analytical Contributions and Strengths

6.1.1. Methodological Contributions

MilieuxVie addresses three methodological gaps identified in the background review. The primary innovation is methodological rather than technical: the integration of dwelling-weighted residential-parcel origins, milieu-adaptive thresholds, and urban-perimeter disaggregation into a single reproducible framework for the rural–urban continuum. First, it operationalises a Québec-specific complete-neighbourhood framework [3] with residential-parcel origin points and dwelling-unit weighting, improving on grid- or centroid-based approaches that dilute signals in municipalities with concentrated settlement patterns. Second, the adaptive radius tiers provide a context-sensitive correction that allows meaningful inter-municipal comparisons across the rural–urban continuum without requiring municipality-specific calibration data. Third, the urban-perimeter analysis mode aligns the tool directly with Québec’s provincial planning framework, making outputs directly usable in GLPG -mandated planning processes.

From a practical standpoint, the tool’s single-file, browser-based architecture is a deliberate design choice aimed at knowledge transfer in resource-constrained institutional contexts. Municipal planners and elected officials in the Laurentides region generally do not have access to GIS software, data servers, or dedicated spatial analysis capacity. A tool that can be distributed by email and opens in a standard web browser substantially lowers the barrier to evidence-based planning at the local scale.

6.1.2. Methodological and Practical Strengths

One methodological advantage of this approach compared to tools that rely on proprietary or static point-of-interest databases is the co-productive nature of its underlying data source. Because service locations are drawn from OpenStreetMap in real time, any errors or omissions identified during an analysis session can be corrected directly in the OSM platform by the user—whether a municipal planner, a public health professional, or an engaged citizen. These corrections are immediately reflected in subsequent analyses, creating a feedback loop between the assessment tool and the geospatial commons. This dynamic is particularly valuable in low-density municipalities where OSM coverage is uneven: the identification of a missing pharmacy or bus stop in the gap analysis output provides a concrete, actionable reason for local stakeholders to contribute to OSM, thereby improving both the local knowledge base and the reliability of future assessments. The tool thus functions not only as an analytical instrument but also as a potential driver of data quality improvement in the communities it serves.

The adoption of urban perimeters as the primary sub-municipal analysis unit constitutes a second important strength. Under Québec’s provincial planning framework (GLPG) [20], urban perimeters represent the designated planning canvas within which municipalities are expected to concentrate new development, densification and service provision. They operate as strategic zones for the planners and urbanists: the bounded territory within which zoning regulations, mixed-use designations, density bonusing, and service infrastructure investments are calibrated. By grounding the completeness assessment in these perimeters rather than in full municipal boundaries, MilieuxVie produces gap analysis outputs that map directly onto the planning levers available to local decision-makers. A measured gap is a prompt for inquiry rather than a prescription: low proximity for a service may reflect demographic, market or regulatory factors the model does not observe, so the tool is best positioned as a screening and prioritisation aid feeding planning deliberation, not a direct map from measurement to intervention. A municipality that learns its urban perimeter scores poorly on healthcare proximity or cycling infrastructure can translate that finding immediately into zoning amendments, land acquisition priorities, or capital infrastructure submissions, without needing to reconcile the result with the performance of its vast uninhabited hinterland.

The use of straight-line (Euclidean) distances, while a methodological approximation discussed below among the tool’s limitations, confers a significant computational performance advantage that has direct implications for user experience and adoption. Network-based routing calculations require server infrastructure, pre-processed routing graphs, and substantially longer computation times at regional scale. The Haversine distance approach, by contrast, enables the scoring of all 93 territorial units or 92 urban perimeters entirely within the user’s browser, in approximately two to three minutes end-to-end, the large majority of which is the initial regional Overpass query. This matters because research on web application usability consistently finds that user abandonment rates increase sharply with page load and computation times: Nielsen [56] established that attention and engagement begin to erode after approximately ten seconds of wait time, a threshold corroborated by subsequent studies on interactive mapping applications [57]. By design, the tool aims to return regional-scale results quickly enough to be usable in time-constrained settings such as council meetings or planning consultations, rather than requiring the five to fifteen minutes of a sequential workflow. This is a design intention rather than a demonstrated outcome: the tool’s usability and uptake in planning practice were not formally evaluated, and a usability study with planners is identified as a priority for future work. The computational efficiency of the straight-line approach is therefore not merely a technical convenience but a deliberate design trade-off in favor of accessibility and usability. To quantify the bias introduced by straight-line distances, we compared Haversine distances against pedestrian network distances (OSRM foot routing) for a stratified random sample of ten municipalities (three dense, four intermediate, three rural; n = 852 origin–facility pairs), computing the network distance to each origin’s straight-line-nearest facility in every category. Because the straight-line-nearest facility is not necessarily the network-nearest one, the measured detour factors are upper bounds on the true detour to the network-nearest facility; the validation is therefore conservative with respect to the tool’s accuracy. Network distances exceeded straight-line distances by a median factor of 1.43 (interquartile range 1.22–1.80), a median detour of roughly 43%, rising from dense (1.39) through intermediate (1.44) to rural (1.76) milieu, where sparse pedestrian networks and water crossings generate occasional very large detours (the absolute-error distribution is consequently right-skewed, so median statistics are reported). Critically, the rank ordering of accessibility was preserved (Spearman ρ = 0.91 overall; 0.98, 0.80 and 0.79 by milieu), and only 7.6% of within-radius classifications changed when network distances were substituted (65 of 852 pairs), concentrated in the dense and intermediate milieu (7.6% and 10.2%), with no change in the rural sample, whose generous radii absorb the detour. Straight-line distances therefore systematically and predictably underestimate travel distance—absolute scores are optimistic—but the relative accessibility picture and the milieu comparison on which the tool’s conclusions rest are robust to the choice of distance metric (Supplementary Table S4).

We further assessed the completeness of OpenStreetMap, the tool’s service data source, against the property roll assessment (2026) with a non-residential property inventory for a stratified sample of 15 municipalities (Supplementary Table S5). Coverage was high for the categories anchoring the essential dimensions of the service- and mobility-proximity (SMP) score, food retail (82%), pharmacies (83%), primary schools (95%) and secondary schools (100% detection of the reference establishments), but lower for child-care services (28%) and health facilities (14%); leisure and natural spaces diverged mainly in granularity, with OpenStreetMap typically recording many more individual features than the corresponding reference classes. Accessibility to child-care and health services may therefore be understated where OpenStreetMap coverage is sparse, a limitation that reinforces the tool’s intended use as a screening and diagnostic instrument rather than a definitive accessibility measure.

6.2. Limitations

The near-universal “incomplete” result should be read with caution: it may reflect the calibration of the framework, the thresholds and the 70% target, as much as genuine territorial deficit. The sensitivity analysis below examines this directly by reporting how the classification distribution responds to alternative targets and density cut-offs.

Before turning to limitations, we note that the tool’s headline outputs are robust to its principal design choices, as confirmed by three sensitivity analyses on the 78 scored units (Supplementary Table S6). First, re-weighting the composite to double the weight of essential services (food retail, primary healthcare, childcare, pharmacy and schools) leaves the ordering essentially unchanged relative to the equal-weight composite (Spearman ρ = 0.99), shifting scores by a mean of 1.3 points and changing the completeness band of only one unit. Second, varying the density cut-offs that define the milieu classes by ±20% (i.e., 8 and 80, or 12 and 120 dwellings/km²) reclassifies at most 5 of the 78 units under either a 20% decrease or a 20% increase in the cut-offs. Third, the finding that no unit reaches the completeness target is insensitive to the target level: moving the 70% threshold to 60% or 80% yields at most one “complete” unit. These analyses indicate that the rural–urban gap and the region-wide shortfall reported above are not artefacts of the weighting scheme, the density cut-offs, or the chosen target level.

Several limitations should be noted. First, the use of Euclidean (straight-line) distances rather than network distances may overestimate accessibility in municipalities where water bodies, rail lines or expressways create barriers between parcels and nearby services. The direction and magnitude of this bias will vary by municipality and service type. As discussed in Section 6.1.2, however, the straight-line approximation is a deliberate design trade-off that enables near-real-time browser-based computation across the full study region, and its practical consequences are attenuated in Québec’s small urban centres where street layouts are relatively grid-like.

Second, OSM completeness varies substantially across the study area. Service facilities that exist physically but are absent from OSM are counted as absent in the analysis, producing artificially low scores for poorly mapped territories. This is particularly likely for small-scale local food retailers, informal childcare arrangements, and community centres in small northern municipalities. Future versions of the tool should incorporate a completeness indicator (e.g., OSM building-footprint density as a proxy for mapping effort) to contextualise scores.

Third, basing the milieu-type density on the residentially designated urban area rather than the total municipal territory is an approximation. Because the denominator captures only the land designated for residential development, a predominantly rural municipality with a small, compact designated core can be classified as dense, and the classification is sensitive to how tightly each MRC has drawn its urban designations. Integration of population density from the 2021 census dissemination areas would provide a more directly comparable measure, at the cost of additional data processing and a more complex user experience.

Fourth, a significant but often overlooked source of measurement error in northern municipalities is the prevalence of secondary and seasonal dwellings. In many Laurentides municipalities north of the Laurentian Mountains, a substantial proportion of residential parcels are used as seasonal cottages whose occupants are not permanent residents. Including these parcels in the accessibility calculation inflates the apparent deficit for services such as childcare and pharmacy, producing scores that underestimate the level of service proximity available to full-time residents. Future versions should distinguish permanent from seasonal residential use using additional occupancy data.

Fifth, concerning spatial measurement, areal and linear OSM features (parks, forests, cycleways) are currently represented by the bounding-box centre returned by the Overpass “out center” option. Consequently, distances to them are measured to that center rather than to the nearest edge, biasing greenspace and cycling coverage conservatively (downward). A future version will use full geometries with nearest-point computation.

Sixth, the tool’s reliance on OSM for green and natural space detection systematically underestimates access to nature in northern municipalities, where vast forests, lakes, and natural terrain surround residential parcels but are not labelled in OSM as parks or leisure areas. A complementary raster-based approach using satellite-derived greenness indices (NDVI) could address this in a future version.

Seventh, the four housing-dimension parameters (dwelling-size diversity, rental proportion, vacancy rate, long-term affordability) are not yet integrated into the composite score. These dimensions are central to Vivre en Ville’s complete-neighbourhood concept, and their exclusion means the composite score currently captures only service proximity and mobility. Incorporating tenure and typology data from the 2021 census, and vacancy rates from the CMHC rental market survey, is a priority for future development.

Finally, following Penchansky and Thomas’s (1981) [58] multidimensional framework of access, the tool strictly measures geographic accessibility (spatial proximity) to the nearest mapped facility. It does not capture the remaining dimensions of realised access: service capacity (availability), quality (acceptability), affordability, or operating hours (accommodation). A single pharmacy or childcare facility within range may be geographically accessible but functionally insufficient for the population it nominally serves. Proximity is therefore a necessary but not sufficient condition for effective access, and the scores should be interpreted purely as a geographic proximity screen rather than a comprehensive measure of realised access [58].

6.3. Generalisability and Future Directions

The methodological framework presented here is specific to the Laurentides region in its data inputs (provincial property roll, Québec urban perimeter layer, Vivre en Ville thresholds) but is generalizable to other Québec health regions with minimal adaptation. Extension to other Canadian provinces or other countries would require substituting equivalent residential parcel datasets and adapting the service-category definitions to local planning frameworks.

The gap analysis module, which identifies priority service deficits at the municipal and urban-perimeter levels, has practical potential as a planning support tool for GLPG implementation reviews, municipal master-plan updates, and health-impact assessment scoping exercises. However, a measured gap is a prompt for inquiry, not a prescription: low childcare proximity may stem from demographic, market or regulatory factors that the model does not observe. The tool is best positioned as a screening and prioritisation aid feeding into planning deliberation, not as a direct map from measurement to policy action.

Future development priorities include: (1) integration of network-distance calculation via the OSRM or Valhalla routing engines; (2) automatic OSM completeness flagging; (3) full integration of census housing parameters into the composite score; and (4) automated report generation in institutional format for direct use in public health knowledge-transfer communications.

7. Conclusions

This paper presented MilieuxVie, a browser-based open-source tool for assessing complete-neighbourhood accessibility across the Laurentides health region of Québec. By combining the Vivre en Ville (2026) [3] 15-parameter framework with real-time OpenStreetMap queries via the Overpass API, provincial residential parcel data from the MAMH property roll, adaptive distance thresholds calibrated to three territorial types, and sub-municipal urban-perimeter analysis, the tool provides a level of methodological rigor and territorial sensitivity that is absent from generic walkability platforms, while remaining deployable in planning offices without GIS infrastructure.

The application of the tool to the 93 territorial units of the Laurentides region (76 municipalities, 11 land-based unorganized territories, 2 indigenous territories and 4 aquatic units) produced three main empirical findings. First, the overall level of complete-neighbourhood accessibility is low throughout the region: the median composite service- and mobility-proximity (SMP) score was 15% across 78 scored territorial units (mean 20.2%; Kanesatake Mohawk Territory is excluded from the municipal statistics), and no municipality reached the 70% completeness threshold. Scores differed significantly across the three milieu types (dense, intermediate, rural; Kruskal–Wallis p = 0.006): the adaptive radius tiers narrow but do not eliminate the structural rural–urban gap, which remains statistically significant. A marked north–south gradient was observed: the southernmost MRC (Thérèse-De Blainville) recorded a median score of 44%, while Mirabel scored 32%, while the northern MRCs of Rivière-du-Nord and Argenteuil recorded medians of 10% and 5%.

Second, the parameter-level analysis reveals a consistent hierarchy of accessibility gaps. Food retail and the cycling network were the best-performing dimensions, while childcare, shared mobility, and pharmacy showed near-universal deficits (all 78 units scored below the 70% target for childcare and pharmacy). These findings have direct policy implications: childcare and pharmacy access are provincial health-equity priorities in Québec; given the low OSM detection rates for childcare (28%) and health facilities (14%), part of the measured deficit may reflect data incompleteness rather than true service absence, and the systematic gaps flagged here should therefore be read as identifying priority areas for further verification and planning investigation in support of GLPG-aligned planning amendments.

Third, the urban perimeter analysis substantially improved on the municipal-scale picture. Among 89 scored perimeters (covering 161,544 dwelling units), two reached the 70% threshold (Sainte-Adèle and Mont-Tremblant, both at 74%) and the regional median rose from 15% (municipal) to 26% (perimeter). The case studies of Sainte-Adèle (74%, four gaps) and Mont-Tremblant (74%, five gaps) illustrate how sub-municipal disaggregation transforms an intermediate-level municipal score into an actionable planning brief: for each perimeter, the gap analysis identifies the specific service types falling below target, ranks them by deficit magnitude, and implicitly points to the planning instruments (zoning, facility siting, transit investment) most likely to close each gap.

Beyond its regional application, MilieuxVie illustrates a broader principle: that open data (OpenStreetMap(OSM)), open government data (provincial property rolls, planning boundaries), and open-source web mapping can be combined into planning-support tools that serve knowledge transfer to non-specialist audiences at virtually no infrastructure cost. A distinctive feature of the OSM-based approach is that errors or omissions identified during an analysis session become direct contributions to the geospatial commons: a planner who notices a missing pharmacy in the gap analysis output can add it to OSM, improving the tool’s reliability for the entire community. The tool’s co-productive data model thus creates a virtuous cycle between territorial assessment and crowd-sourced data quality improvement.

Future development priorities include: (1) integration of network-distance routing via OSRM or Valhalla to replace straight-line distance approximations; (2) full integration of the four housing-dimension parameters (diversity, tenure, vacancy, affordability) from Statistics Canada and the SCHL; (3) automatic OSM completeness flagging to contextualise low scores in poorly mapped municipalities; and (4) automated report generation in institutional format for direct use in public health communications. As Québec municipalities undertake Government land-use planning guidelines (GLPG)-mandated planning revisions over the next decade, tools of this kind may play a growing role in anchoring land-use and infrastructure decisions in population-level accessibility evidence.