Promoting Public Health Through Urban Walkability: A GIS-Based Assessment Approach, Experienced in Milan

Abstract

Introduction. The global challenge of physical inactivity necessitates innovative approaches and strategies to optimize built environments in order to promote healthy and sustainable lifestyles, such as active transportation. For this reason, walkability is a crucial area of research in urban health, with several studies focusing on assessment frameworks. However, a gap persists between theoretical development and practical implementation. This study explores the application of the Milan Walkability Measurement Tool (MWM-Tool), a walkability assessment framework previously developed by Politecnico di Milano, to evaluate the urban features in favor of walkability by integrating GIS technology with an extended testing scope. It is based on a scientific approach utilizing 10 sub-indicators divided into three macro-areas (Density, Diversity, Design), identified through a comprehensive literature review. Method. Focusing on the application of the MWM-Tool in Milan, the study employs the 88 Nuclei of Local Identity (NILs), which are the official designations for Milan’s neighborhoods, as the units of urban analysis. Based on previous experience, the digitalization of the assessment framework has been improved: geospatial data corresponding to 10 sub-indicators were filtered to generate vector layers, primarily sourced from two public geographical platforms. The GIS-based method produces thematic maps evaluating all neighborhoods according to the dimensions of Density, Diversity, and Design. Darker and lighter colors represent the range of the scores. Both single indicators and macro-area maps, as well as overall walkability level maps, were generated to illustrate the results. Result. The results of the macro dimension assessment, combining 10 sub-indicators, provide an objective view of the distribution of walkable space quality in Milan. Only 7 out of 88 neighborhoods achieved the highest score, all of which are located in the city center, while suburban areas showed significantly lower scores. By incorporating census GIS data, the study also identified the population distribution across areas with varying walkability levels. Based on the results of the assessment, it may be possible to develop and prioritize the optimization of walkable features, revitalizing underserved areas and fostering a healthier community environment. Conclusion. The georeferenced-data maps represent an effective tool to highlight both neighborhoods with high urban quality, which could be used to promote active mobility and healthy lifestyle adoption, as well as those requiring improvement strategies from policy and decision makers. The research output provides a reference for further urban planning initiatives in Milan and contributes to enhancing pedestrian-oriented built environments. Using GIS open-source data, the method is scalable and can be easily replicated in other cities. It could also be used as a system for monitoring walkability over time.

1. Introduction and Theoretical Scenario

1.1. Public Health, Well-Being and Urban Environment

Physical inactivity is now identified as the fourth leading risk factor for global mortality [1,2]. Physical inactivity levels are rising in many countries, with major implications for the prevalence of non-communicable diseases (NCDs) and the general health of the population worldwide [3]. As a crucial public health issue, insufficient physical activity (PA) is highly related to technology, economic factors, and the built environment, increasing after the pandemic period [4,5]. Technology offers a variety of means [6] by which to significantly reduce the energy consumption required for activities in daily life. At the same time, economic development has shaped a model of a fast-moving society [7] where high salaries encourage people to choose more sedentary jobs over more active ones. The built environment, on the other hand, can influence mobility choices [8,9] especially in urban contexts. Rapidly increasing obesity rates are a negative factor for public health, reflecting a lack of energy balance [10]. PA is the only flexible component of daily energy expenditure to replace sedentary routines. For example, opting for walking or cycling instead of driving for short trips allows individuals to expend additional calories while covering the same distance, thereby promoting PA and providing overall health benefits [11].

In adults aged 18–64 years, PA includes leisure time, transportation opportunities (e.g., walking or cycling), occupational (i.e., work) movement, sport, or planned exercise, in the context of daily, family, and community activities [11,12]. The World Health Organization (WHO) recommends that adults aged 18–64 years should do at least 150 min of moderate-intensity aerobic PA throughout the week, at least 75 min of vigorous-intensity aerobic PA throughout the week, or an equivalent combination of moderate and vigorous-intensity activity [13]. The concept of accumulation refers to achieving the recommended 150 min of PA per week by engaging in multiple, shorter bouts of at least 10 min each, distributed throughout the week. The total duration is then calculated by summing the time spent in each session. For example, this could be achieved through 30 min of moderate-intensity activity, five times per week. Evidence in the form of biomedical markers [14] underlines the benefits of undertaking regular PA throughout the week (e.g., five or more times per week). Moreover, incorporating PA into daily routines—such as through active travel modes like walking and cycling—can enhance long-term adherence to an active lifestyle [15]. For the reasons cited above, the urban context [16,17] and the walkable environment play an important role in fulfilling the WHO’s recommendations, since it could encourage people to be more active by its design features [18,19,20].

Compared to other modes of transportation, walking and cycling play a minor role, as private cars are generally considered a more comfortable option for most trips. Walking and cycling, in fact, are more commonly used as forms of PA rather than primary modes of travel [21,22]. Compared to other modes of exercise, they are popular because they are relatively cheaper and offer relatively little risk of injury [23].

According to a study about the epidemiology of walking for PA in the United States [15], since walking can be done with relative ease for most people and can be done at a multitude of speeds and intensities and over different distances, it seems to be the ideal PA for someone who is in less than favorable health. Additionally, developing and urban environments can encourage active transportation choices [24], thus promoting walking and cycling as a way to improve health. Some observational studies [25] indicate that responses to the question of whether individuals have changed their walking behavior since they began utilizing a particular new walking trail provide valuable feedback on the behavioral changes associated with urban environmental aspects. Indeed, 45% of regular walkers and 50% of occasional pedestrian stated that they walk mostly within their own neighborhood [26]. The study data also show that over 60% of the respondents who were either occasional or regular walkers reported using neighborhood streets as their main location for walking for PA, and 33.6% of the surveyed population reported that they attain the recommended levels of PA by walking.

1.2. Physical Activity and Urban Environment

According to the WHO, low or decreasing PA levels often correspond with a car-oriented urban environments. In fact, increased reliance on passive daily modes of transportation further contributes to insufficient PA among the population. When shifting the perspective to the urban context, a critical question is whether the existing built environment adequately encourages people to choose active transportation modes [27]. To understand urban travel behavior, it is essential to examine the built environment [28] at the neighborhood scale. Compared with car-oriented trips, walking trips are more strongly influenced by neighborhood characteristics than by the structure of the whole city. Addressing this issue requires a comprehensive assessment of the urban environment [29], integrating detailed data on both the PA and the built environment feature (promoting walkability and cyclability) levels. Regardless of the features considered, data on walking behavior must be spatially correlated with detailed data on the built environment.

Several questions remain about the interconnectedness of built environments, travel behavior, and public health [30]. To prioritize health benefits, increasing the share of walking and cycling can be achieved through multiple strategies [31]. Efforts to increase pedestrian orientation in a built environment through mixed use development, street connectivity, and high-quality urban design can enhance both the feasibility and the attractiveness of walking and bicycling by mitigating physical [24] and psychological barriers [32]. Even a modest increase in walking would help to substantially improve the health and quality of life of most people. The contemporary challenge is to understand and evaluate the relations between the built environment and human behavior and then to develop models that can predict the environmental conditions under which humans will be more physically active. Such models can assist planners in the design and policy makers in the management of built environments to promote PA. The available evidence lends itself to the argument that a combination of urban design, land use patterns, and transportation systems that promotes walking and cycling will help the creation of active, healthier, and more livable communities. Advancing the development of healthier and more livable communities requires a collaborative research approach that integrates the research paradigms of both urban planning and public health. Such interdisciplinary approaches are essential for fostering innovative solutions that enhance the construction of environments that promote public well-being.

2. Materials and Methods

This research was based on the development of a method to apply the MWM-Tool at a macroscale level to an entire urban context using open-source geo-localized big data. For this purpose, the method included several phases: MWM-Tool analysis at the macroscale level, identifying the test area and spatial data collection, and data processing techniques.

2.1. Evaluation Framework for the Walkability: MWM-Tool

Regarding collaborative experiences, a previous study conducted by the authors defined the Milan Walkability Measurement (MWM) Tool [33] based on a scientific approach that integrates perceived tools [34], observational tools, and archival tools. The MWM-Tool arose from a comparison and critical selection of various qualitative–quantitative indicators that were integrated into the multi-criteria analysis structure of a dual-scale survey [29] with reference to walkability and to indicators that have implications for health promotion. The MWM-Tool operates on both the macro dimensional (urban area) and micro dimensional (street level) spatial scales. The macro dimension (Density, Diversity, and Design criteria) refers to the urban scale and examines the city from above to identify which neighborhoods are more critical or more inviting in terms of walkability. It quantitatively describes urban factors which can influence the propensity to walk in each neighborhood. The micro dimension (Usefulness, Safeness, Comfort, and Aesthetic criteria) investigates the city at the street scale. By adopting observational data, this level of analysis provides qualitative assessments of outdoor spaces, focusing on street segments ranging from 500 to 700 m in length. Such analyses are fundamental to identifying design and functional actions that could improve walkability in the redesigning of public spaces or in the urban regeneration of a given area.

The assessment framework is structured around seven different dimensions (three macro dimensions and four micro dimensions) and multiple sub-indications, which were weighted by a panel of experts. Based on a previous assessment framework definition [33], a panel of 12 experts, including professionals in urban planning, transportation, architecture, and public health, was assembled to assess the relative importance of each walkability dimension and its sub-indicators. Experts performed pairwise comparisons, using Saaty’s 1–9 scale, providing a systematic way to assess trade-offs between different walkability attributes. The priority weights were derived using the Eigenvector Weight Computation method, a mathematically robust and consistent weighting system. The reliability of the expert judgments was validated through a consistency ratio (CR) check, with adjustments made if the CR exceeded 0.1 to enhance coherence in the weighting assignments. Finally, to ensure weight integration in our walkability assessments, the final dimension and sub-indicator weights were incorporated into the walkability score computation, allowing us to perform transparent and replicable assessments of urban walkability across different areas. By integrating AHP-based weighting, this methodology ensured that the MWM-Tool was both data-driven and expert-informed, enhancing the credibility of the walkability index and enabling the authors to devise evidence-based urban policy recommendations.

This framework introduces innovative criteria, including a multi-scalar assessment approach (macro and micro dimensions) and the integration of health-related factors linked to the promotion of active transportation and healthy lifestyles, as identified in the literature. The design recommendations, based on the collected qualitative and quantitative data, provide valuable insights for policymakers to make evidence-based decisions [35,36]. Likewise, this approach supports urban designers in understanding what aspects of an urban environment must be improved or implemented to promote more walkable and health-supportive cities. The MWM-Tool may be used as supportive assessment framework; it is capable of rapidly delivering recommendations, thereby supporting the decision making process regarding urban planning strategies.

The macro and micro dimensions can be used together as two levels of walkability analysis or individually—one to assess an entire city (highlighting more or less critical neighborhoods for targeted quantitative strategies) and the other to examine a specific urban area in detail, guiding precise architectural and spatial planning.

The main difficulties in applying the MWM-Tool at the macroscale level are the availability and objective systematization of all the information required for the evaluation. For this reason, the aim of the present research is to develop a GIS-based method capable of rapidly generating thematic maps representing the MWM-Tool at the macroscale level. The use of GIS-based data makes it possible to rapidly and objectively evaluate neighborhoods, underlining strengths, weaknesses, and priorities to support stakeholders (policy and decision makers) to guide urban policies in favor of walkability promotion. The present research is the first such first pilot test in Milan but could be transferred and repeated in other cities that are sensitive to the promotion of healthy lifestyles.

2.2. Milan Walkable Measurement Tool (Macro Level)

As noted before, the MWM-Tool has a deep theoretical research background, since it is based on a summary and innovation of 20 existing walkability assessment tools. Although its effectiveness has been demonstrated in another study by a test on small scale, it was necessary to extend the scope of the test to the whole city of Milan and to exploit its application potential. While the micro analysis was observational and an accurate evaluation was quite arduous, the macro analysis was archival and could be conducted digitally. The macro scale is quantitatively related to urban planning (morphological aspects, presence of services and infrastructures) and may be divided into three dimensions with 10 sub-indicators (Figure 1).

The first dimension, titled Density, quantitatively analyzes the degree of availability and investigates the presence/absence of urban features that are directly related to the promotion of walkability. The first sub-indicator of Density analyzes Intersections (1.1), i.e., it is related to the number of intersections present within the analyzed area. The second sub-indicator analyzes Built-up areas (1.2), that is, the ratio between built-up areas and the total analyzed land area. The third sub-indicator analyzes Destinations (1.3) and is related to the number, typology, and positions of useful functions and amenities placed at the ground level. The fourth and last sub-indicator analyzes the Presence of Sidewalks/Cycling Lanes (1.4); it indicates the presence of cycling lanes and sidewalks on one side or both sides of the streets within the analyzed area.

The second dimension, titled Diversity, analyzes the infrastructure system of an area in terms of typologies of transportation. The first sub-indicator of Diversity, Transportation Stops (2.1), indicates the availability of transport stops within the analyzed area, including the typology of public transport. The second sub-indicator, Coverage (2.2), is related to the area covered by a radius of 300 m from public transport stops which correspond to 10 min of walking for people with disabilities and 5 min of walking for people without. The third sub-indicator, Parking Availability (2.3), is an indication of linear meters of on-street parking within the area. A large availability of parking lots is not supportive of walkability within a neighborhood, instead indicating a car-oriented urban environment.

The third dimension, titled Design, analyzes the urban morphology and design, including the street hierarchy, the built fabric typology, and the presence of greenery along the streets. The first sub-indicator of Design, Street Layout (3.1), is related to the street design in terms of safety and speed limits. The second sub-indicator, Building Layout (3.2), analyzes the areas consisting of a recognized urban design. The third sub-indicator of Design, Green Layout (3.3), evaluates the presence of streets with greenery along most of their length (Figure 2).

Calculation of the MWM-Tool Macro Scale:

Den. = In. × 0.15 + Bu. × 0.1 + De. × 0.3 + Sc. × 0.45,

(1)

Div. = Ts. × 0.36 + Tc. × 0.49 + Pa. × 0.15,

(2)

Des. = Sl. × 0.61 + Bl. × 0.15 + Gl. × 0.24,

(3)

MWM = Den. × 0.16 + Div. × 0.24 + Des. × 0.6

(4)

Using the pairwise correlation method, thanks of the experts involved, a specific weight was assigned (Figure 1b) to each sub-indicator and to each macro dimension (0.16 for Density, 0.24 for Diversity, and 0.6 for Design). The assessment results of the sub-indicators were calculated according to each scoring method. The three dimensions were calculated by summing the results of multiplying the assessment results of sub-indicators by each sub-weighting. In this way, a value representing the degree of walkability for Milan was obtained.

2.3. Defining the Testing Area and Spatial Data Collection

The expanded test area is the whole city of Milan [34], as it could provide a reference for pedestrian-oriented urban planning in other cities globally. In our study, zoning the city in a rational way was necessary. The Milanese Nuclei of Local Identity (NILs) represents the best territorial atlas for use as a reference to verify the planning of services but also for knowledge about the neighborhoods that make up the various local settings, highlighting unique and different characteristics for each. Such a principle of territorial division and a model of dynamic management are logically consistent with the requirements of this study, so the NILs were used directly. NILs are capable of categorizing as “urban clusters” the main geographical, demographical, socio-economic, and urban features data. However, it should be kept in mind that borders are porous. The present authors considered NILs as the primary geographical sub-divisions for the city of Milan, considering that urban policies are equally planned according to this reference.

Two platforms were used to complete the collection of geospatial data: Milan Geoportale (The Geoportal of the Municipality of Milan); and OpenStreetMap (OSM). By filtering and analyzing the data packets, the vector layers corresponding to all evaluation indicators could be obtained directly or indirectly, as well as the boundary and segmentation layers supporting our analysis. All the indicator-related data were obtained by direct selection, manual sorting, or automatic calculation (Figure 3).

2.4. Crucial Data Processing Techniques

Based on our previous MWM-Tool experience, one of the crucial tasks of the present research was to improve our geometric and mathematical evaluation of the three dimensions and ten sub-indicators to obtain a systemic and global assessment of the urban area. The vector spatial data layers corresponding to the evaluation indicators could be divided into three categories according to their data features. There were four common critical operations when performing vector calculations (

Table 1. Critical operations for three types of vector layers.

	Type	Icon	Corresponding Sub-Indicators	Typical Operations
					Merged Attributes		Breakdown to NILs		Merge Layers		Attribute Calculation
1	Point		In. De. Ts.
2	Line		Sc. Pa. Sl. Gl.
3	Polygon		Bu. Tc. Bl.

). In the first step of dissolving attributes, the layer with vector elements often had complex attributes which did not match the attribute that had to be limited in each NIL. Therefore, it was necessary to dissolve all the vector elements in the same layer by using the command “Dissolve”. In the second step, i.e., breakdown to NILs, using the command “Clip” helped to break down the pure vector layer obtained in the last step. With the help of the NIL sheets, the command “Clip” was applied 88 times. In this approach, all the vector elements were reorganized to each NIL, and 88 new layers were created. In the third step, i.e., merge layers, the command “Merge vector layers” helped to automatically calculate the data in the attribute table. The results were all the vector elements in the new layer that was selected in their respective NIL intervals. In the fourth step, i.e., attribute calculation, we opened the attributes table and created a new attribute column to automatically calculate the data using the point count, length count, and area count formulas.

The data obtained directly were usually not the final target; rather, further calculations were done automatically by creating a new attribute column in the attribute table and inserting a formula. The processing of the spatial data involved in this study was complex and the calculation of each sub-indicator presented certain peculiarities that needed to be distinguished separately. These issues are described in the Further Testing section of the specific evaluation metrics. The application and validation phase of the tool for the city of Milan and the results and validation of the GIS-based method are reported in the following section.

3. Application Results of the MWM-Tool (Macro Level) in Milan, Italy

Based on each corresponding scoring method, 10 workflow charts were developed for all 10 sub-indicators to guide the practical operation of the assessment progress. After completing the data processing, the assessment results were graphically illustrated and a corresponding summary statistical analysis was conducted. The geospatial data layers were classified into three types, i.e., point vector layer, line vector layer, and polygon vector layer. The technical operations of three sub-indicators were based on the point vector layers Intersections, Destinations, and Transportation Stops. The technical operations of four sub-indicators were based on line vector layers which included Sidewalk/Cycling Lanes, Parking Availability, Street Layout, and Green Layout. The technical operations of three sub-indicators were based on polygon vector layers, i.e., Built-up Area, Coverage, and Building Layout. In view of the specificity and difficulty of the testing process for each assessment indicator, Destinations, Sidewalk/Cycling Lane, and Coverage were selected as being representative of the different types of vector layers for our overview interpretation.

3.1. Representative Sub-Indicator as a Point: The Example of Destinations

Considering the huge data classification task and repetitive calculations, the sub-indicator Destinations was chosen as the most representative one. The specificity of this indicator lies in ensuring the completeness and timeliness of the data and the accuracy of the data classification. A positive link between destination diversity and walkability was confirmed, which is further evidence that the features of the built environment are important in the design of neighborhoods to increase walking and meet the health needs of residents. As defined by the researcher, Destinations comprised 10 categories: Culture, Religion, Active, Medical, Eating, Market, Trading, Nightlife, Shops, and Hotel. (

Table 2. 10 categories of destinations.

	Types	Category of Destinations	Number	Density/km2
1	Culture	Memorial, archaeological, kindergarten, monument, artwork, ruins, chemist, tower, university, library, college, school, museum, arts center, theatre, shower, recycling, planetarium, inscription	2951	16.23
2	Religion	Place of worship, church	363	2
3	Active	Pitch, picnic site, garden center, viewpoint, swimming pool, bench, park, drinking water, playground, water well, attraction, fountain, sports center, biergarten, theatre, dog park	12,219	67.21
4	Medical	Doctors, optician, dentist, hospital, pharmacy, shelter, nursing home, orthopedics, clinic	1196	6.58
5	Eating	Kiosk, bakery, restaurant, café, bar, beverages, pub, fast food, cheese, chocolate, ice-cream, Juice, seafood, tea, wine	9477	52.13
6	Market	Supermarket, market, greengrocer, mall	581	3.20
7	Trading	Community center, comms tower, town hall, courthouse, post office, post box, police, bank, tourist info, charity, coworking space, insurance, prison, social center	2475	13.61
8	Night life	Nightclub, gambling	50	0.28
9	Shops	Vending parking, do-it-yourself, florist, veterinarian, car sharing, outdoor shop, cinema, bicycle rental, computer shop, car rental, laundromat, car wash, jewelers, recycling paper, gift shop, sports shop, travel agent, video shop, newsagent, atm, car dealership, telephone, camera surveillance, water tower, vending machine, beauty shop, butcher, cigarette vendor, bicycle shop, hairdresser, mobile phone shop, convenience, stationery, recycling clothes, toy shop, bookshop, furniture shop, greengrocer, fire station, waste disposal, shoe shop, toilet, recycling, department store	11,276	62.02
10	Hotel	Guesthouse, hotel, motel, hostel	818	4.50

).

3.1.1. Scoring Method

Destination density (De) was scored by comparing the density of destinations in each NIL (Dex.NIL) with the average density of destinations in Milan (De.Milan). Given the various types of destinations, such a comparison had to be very precise. The density of each type of destination in each NIL (Dex.NIL) was compared with the average density of each type in Milan (Dex.Milan) one by one. If the density of a type of destination in a NIL was bigger than the average density of that type in Milan, the NIL was scored as 0.5. When the density of a type of destination in a NIL was less than the average density of that type in Milan, the NIL was scored as 0. The final score was calculated by summing the scores of different destination types.

Calculation of De.L:

If Dex.NIL ≥ Dex.Milan Then Dex._L = 0.5,

(5)

If Dex.NIL < Dex.Milan Then Dex._L = 0,

(6)

De.L = De1._L + De2._L + … + Dex._L

(7)

3.1.2. Scoring Process

The scoring process consisted of five steps. The first step was vector data classification. There were 10 categories of destinations, and each category had its own destination sub-types. The original point layer of destinations had to be separated into layers of sub-types. Then, according to the category classification, we merged all the vector layers into ten layers for different types of destinations. The second step was to calculate the average density of each type of destination in Milan. The total number of each type of destination in Milan was checked. The area of Milan is 181.8 Km²; therefore, the average density of each type in Milan could be calculated. The third step was to calculate the density of each type of destination in each NIL. These 10 types of the destination layers were typically point layers, so this step could be done by repeating the four critical operations ten times plus performing the density calculation in the attribute table. The fourth step was to score the density of each type of destination for each NIL. The fifth step was to sum the density of each type of destination for each NIL.

3.2. Representative Sub-Indicator as a Line/Surface: The Example of Sidewalk/Cycling Lanes

The indicator of Sidewalk/Cycling Lanes showed a direct correlation between the willingness of public to choose active transportation and urban walkability. The researcher divided the presence of sidewalk/cycling lanes into four levels: sidewalks on both sides with cycling lanes, sidewalks on both sides, a sidewalk on one side, and no sidewalks or cycling lanes. The higher the density of footpaths/cycleways, the more adequate the walkable urban environment.

3.2.1. Scoring Method

The score for the density of Sidewalk/Cycling Lanes (Sc.) was determined by counting the percentage of each level of Sc in each NIL within the total road network in each NIL. Then, it was necessary to determine the score for each level of Sc with the help of the corresponding assignments and percentages (

Table 3. Sidewalk/Cycling Lane scoring levels.

	Level	Assignment	Percentage	Final Score
1	Sidewalks on both sides and cycling lanes	5	Sc1._PCT	Sc1._L
2	Sidewalks on both sides	3.33	Sc2._PCT	Sc2._L
3	Sidewalk on one side	1.66	Sc3._PCT	Sc3._L
4	No sidewalk or cycling lane	0	Sc4._PCT	Sc4._L

). The final score for each NIL was the aggregate of the scores for the four levels in each NIL.

Calculation on Sc._L:

Sc1._L = Sc1._PCT × 5

(8)

Sc2._L = Sc2._PCT × 3.33

(9)

Sc3._L = Sc3._PCT × 1.66

(10)

Sc4._L = Sc4._PCT × 0

(11)

Sc._L = Sc1._L + Sc2._L + Sc3._L + Sc4._L,

(12)

3.2.2. Scoring Process

The scoring process consisted of four critical steps. The first step was to separate the layers for each level of Sc and to perform a length calculation for each level. The second step was to calculate the percentages for each level in each NIL based on the data of the length for each level and road network in each NIL. The third step was to get the scores for each level in each NIL, i.e., the percentage times the corresponding assignment. The fourth step was to calculate the score in each NIL by summing the scores for each level in each NIL.

3.2.3. Crucial Operation Issues

These vector line layers for the four levels could not be obtained directly. There were three vector layers related to the separation that could be obtained directly from the Geoportal of the Municipality of Milan: the polygon vector layer of Pedestrian Circulation Area, the line vector layer of Cycling Lanes, and the line vector layer of Road Network. The vector layer for Level 1 was separated from the vector layer for Cycling Lanes because only locations with cycling paths have the possibility of satisfying Level 1. Moreover, the vector layers for Levels 3 and 4 were separated from the vector layer of Road Network with referencing the Pedestrian Circulation Area polygon vector layer (Figure 4). The vector layer for Level 2 could not be separated from the vector layer of Road Network while the data of the length could be calculated with the help of the data of the length for the other three levels and the Road Network.

The calculation of the length of each grade in each NIL was based on many manual selections, which caused some unavoidable biases. These biases could be reduced by averaging over several iterations but could not be eliminated.

3.3. Representative Sub-Indicator as a Polygon: The Example of Coverage

The coverage of transportation stops is an accurate reflection of the possibility of approaching active transport. A public transport stop within a 300 m radius means that people can reach and use public transport to reach their destination efficiently in only 5–10 min.

3.3.1. Scoring Method

Coverage Diversity (Tc.) was scored by percentage classification of coverage area in each NIL. When the percentage in each NIL was 1 to 20 percent, the NIL was scored as 1. When the percentage in each NIL was 21 to 40 percent, the NIL was scored as 2. When the percentage in each NIL was 41 to 60 percent, the NIL was scored as 3. When the percentage in each NIL was 61 to 80 percent, the NIL was scored as 4. When the percentage in each NIL was 81 to 100 percent, the NIL was scored as 5.

Calculation on Tc._L:

If 0.01 < Tc._PCT ≤ 0.2 Then Tc._L = 1

(13)

If 0.2 < Tc._PCT ≤ 0.4 Then Tc._L = 2

(14)

If 0.4 < Tc._PCT ≤ 0.6 Then Tc._L = 3

(15)

If 0.6 < Tc._PCT ≤ 0.8 Then Tc._L = 4

(16)

If 0.8 < Tc._PCT ≤ 1 Then Tc._L = 5

(17)

3.3.2. Scoring Process

The scoring process consisted of four key steps. The first step was to calculate 300 m isochrones for each type of transportation stop using the vector layers obtained in the first step in the assessment process of Transportation Stops. The second step was to create a coverage layer with the five polygon vector layers of 300 m isochrones obtained in the previous step. The third step was to calculate the percentage of coverage for each NIL. The coverage layer obtained in the last step was typically a polygon layer with no other layers to interact with, so this step could be undertaken by repeating the four critical operations for the polygon layer plus the percentage calculation in the attribute table. The fourth step was to complete the percentage classification to get the score for each NIL. The assessment result of coverage was then obtained.

3.3.3. Crucial Operation Issues

The specialty of coverage is the automatic calculation for the transformation from point vector layers to polygon vector layers. With the help of the ORS plugin in QGIS, the polygon vector layer of the 300 m isochrones for Bike Sharing Spots was calculated from the corresponding point vector layer (Figure 5). According to the same methods, the polygon vector layers for the other four types of transportation stops could then be calculated.

3.4. Walkability Assessment Results

The assessment results of all the sub-indicators were finally calculated, as the challenge of identifying the most representative ones had been overcome. According to the measurement framework, there were four steps that needed to be completed to get the assessment results for Walkability. The first three steps were to calculate the Density, Diversity, and Design in each NIL by summing all the results of the sub-indicators multiplied by each corresponding sub-weighting (Figure 6, Figure 7 and Figure 8). The fourth step was to calculate the walkability in each NIL based on the assessment results of the first three steps (Figure 9).

According to the distribution of the walkability levels, there were only 7 NILs out of 88 with a score between 3 to 4, i.e., representing the highest level of walkability (High walkability); 47 NILs had a Medium-High walkability, while the other 24 NILs were assigned scores between 1 to 2, which correspond to a Low and Medium-Low walkability ratings, respectively. More than 70 NILs were assessed as being below the passing level, which means that there is a high improvement potential for the walkable environment in Milan (

Table 4. Assessment results statistics of Walkability (0 < L < 1 equal to Low walkability; 1 < L < 2 equal to Medium-Low; 2 < L < 3 equal to Medium-High; and 3 < L < 4 equal to High walkability).

Level	Total Number	Detail NILx
1> MWM._L > 0	2	39, 47
2> MWM._L > 1	22	17, 23, 29, 32, 36, 38, 40, 41, 54, 55, 61, 62, 63, 64, 72, 73, 74, 75, 78, 81, 82, 83
3> MWM._L > 2	47	4, 5, 6, 7, 9, 11, 12, 13, 14, 15, 16, 21, 22, 25, 26, 27, 28, 30, 33, 34, 35, 37, 42, 43, 44, 45, 46, 48, 49, 51, 52, 53, 56, 57, 58, 59, 60, 65, 66, 67, 68, 70, 71, 76, 77, 79, 80
4> MWM._L > 3	7	1, 2, 10, 19, 20, 50, 69

, Figure 9). It was not possible to carry out an evaluation on 10 NILs in Milan because they are agricultural or green areas, for which the indicators were not assessable. Furthermore, looking at the ratings obtained in the three dimensions, it is clear that while Diversity (Figure 7) was generally positive due to good public service coverage, the Density (Figure 6) and Design (Figure 8) dimensions were critically low in many NILs.

Depending on the weightings of the four sub-indicators assessed, the trend in the Density dimension scores across the 88 NILs in the city of Milan was strongly influenced by the Sidewalk/Cycling Lanes (1.4) sub-indicator. It was balanced by the influence of the other three sub-indicators, with the final change in scores showing a decreasing trend from the city center toward the peripheral areas. The sub-indicator Destinations (1.3) denoted greater diversity between the city center of Milan and the semi-peripheral neighborhoods, underlining the needs of re-balancing the ratio and better developing those areas in terms of services and amenities.

The change in Diversity scores was generally good; it was strongly influenced by the sub-indicators Transportation Stops (2.1) and Coverage (2.2) of the Public Transport System, that, in Milan, is quite comprehensive. At the same time, the effect of the sub-indicator Parking Availability (2.3) was not significant.

The Design scores were strongly correlated with the Street Layout (3.1) and partially influenced by Building Layout (3.2) and Green Layout (3.3). The distribution of design scores showed a decreasing trend from the city center to the surrounding suburbs, but this phenomenon covered a radial diffusion and did not show meaningful differences between the different parts of the city.

The results of the MWM-Tool assessment show a real picture of the walkable feature distribution in Milan.

As expected, the city center NILs achieved the highest walkability scores, while those on the outskirts recorded the lowest overall scores. Another expected result was the presence of three semi-central areas with low ratings, characterized by former railway yards (NIL-78 and NIL-36) or abandoned industrial sites (NIL-29).

However, some NILs were positive exceptions, as they received the best evaluations despite being located in semi-suburban areas, such as Tortona (NIL-50), Sarpi (NIL-69), and Centrale (NIL-10), or even in Suburban areas (northeast of the city), such as Loreto (NIL-20) and Padova (NIL-19). These NILs are typically marked by a strong historical neighborhood identity, which, in recent years, has increased their attractiveness for both residents and commercial activities. This trend has been further supported by pedestrianization initiatives, limited traffic zones (ZTL), and 30 km/h zones. This confirms the importance of neighborhood identity and slow mobility in terms of enhancing walkability, as emphasized in “15-min city” theories [37].

It is also interesting to note that Milan’s large urban regeneration areas, such as CityLife and Garibaldi, received medium-high but not top ratings. This was partly due to their medium rating of indicators related to speed limits, the presence of cycling infrastructure, and the street-to-building height ratio.

Finally, the majority NILs between the city center and the peripheral neighborhoods showed a high potential for optimizing walkable spatial quality. Based on the results of this study, a series of projects for the optimization of walkable spaces could be introduced to revitalize areas and create healthier community environments (Figure 9).

4. Findings and Discussions

The testing processes were quite straightforward for three sub-indicators, i.e., Intersections, Built-up Area, and Building Layout, as they were basically based on repeating the four critical operations for the vector layer, while the test process for Transportation Stops included an additional step, i.e., separating the point vector layers of five typologies of transportation stops. The calculation of coverage was based on the point vector layer of transportation stops to calculate the isochrones of 300 m radius using the ORS plugin. There was a wide variety of service locations corresponding to destinations and the data were constantly changing over time. Therefore, the frequency of data updates had to be fully considered to ensure the accuracy of the assessment results. The challenge for street layout was centered on the process of converting the original polygon vector layer into the target line vector layer. There were two critical operations for green layout: filter the elements of the original polygon vector layer that belong to the target data, and to calculate the midline of the polygon vector layer using the HCMGIS plugin in QGIS. The target data for Parking Availability was roads with on-street parking. These data are not available to the public and had to be obtained by manual selection from the linear vector data of the road network, drawing upon Google satellite images. The complexity of the test operation of Sidewalk/Cycling Lanes lay in the transformation of two linear vector layers and one polygon vector layer into three linear vector layers which linked to three levels of Sidewalk/Cycling Lanes. According to the application-oriented testing in this research, geo-spatial data acquisition methods for the sub-indicators were found to be more suitable for the data modelling if there was no need for manual selection. Therefore, the original assessment framework [33] could be optimized in the light of this practical experience, increasing the practical value of the MWM-Tool at the macro scale.

Bask to design recommendations, the study presents a comparison among demographic distribution (census data) and walkable features of the urban environment, with the goal of understanding the various factors which promote walkability. The evaluation framework generated a clear image of the current urban scenario, allowing those characters to be in line with the general condition of the studied urban environment [35,38]. Making cities more walkable and cyclable means improving physical factors, like building features, land use mixes, densities, and street design, to create a more convenient, safe, comfortable, and attractive space [39,40] for walking and cycling, especially in the mostcritical NILs.

5. Conclusions and Research Outlook

5.1. Conclusions

Based on the theoretical framework of the WMW-Tool, this research developed rigorous operational workflows for spatial data processing corresponding to each assessment indicator and bridging the gap between the theoretical framework and practical application of an urban built environment assessment [41]. In this process, this research classified and modularized four operational steps into three categories, which effectively reduced the repetitive interpretation of data. In the discussion section, a detailed technical comparison of the operational difficulties encountered in the testing process for each sub-indicator was presented, and an overview of the evaluation results was provided. The application workflow for all other NILs was innovated to help the MWM-Tool play a sustainable role in continuously monitoring the spatial situation. The MWM-Tool has been used, for instance, to support the Municipality of Milan to evaluate the coherence of walkability infrastructure before undertaking physical redevelopment actions. Furthermore, it was used by the authors of this research as a support tool as part of a European project titled “Safely Connected: Sustainable Common Accessibility of Lively Downtowns for Healthy People”, promoted by EIT Urban Mobility “For more livable urban spaces, COVID-19 crisis response”.

This study emphasizes the correlations between the built environment and walkability, evaluating their levels and highlighting the weaknesses that need to be addressed in terms of cyclist and pedestrian accessibility and urban quality. In this way, priority aspects are defined as potential and possible transformations to be undertaken [42]. The relation that exists between urban features and their level of pedestrian-friendliness is evidence-based, and the method can assess both qualitative and quantitative aspects within the complexity of a city environment. Walkable features can directly encourage both active behavior adoption and healthy life-style promotion; these notions, together, represent challenges in the context of global health. This research achieved its purpose of underlining strengths, weaknesses, and priorities for stakeholders (Policy and Decisione-Makers) with the goal of guiding urban policies in favor of walkability promotion. It represents a first preliminary pilot experiment in Milan, to be transferred and repeated in other cities that are sensitive to the promotion of healthy lifestyles.

However, it is important to note that the transfer of the tool to the GIS system involved a series of complex IT processes, and replicating the analysis in other urban contexts would, despite the availability of predefined steps, requirecomparable technical expertise and access to urban data in GIS-supported formats. Finally, it is also necessary to underline that while macro dimensions (urban area, e.g., Density, Diversity, and Design criteria) rely primarily on quantitative data, this study did not explore micro dimension factors (street level, e.g., Usefulness, Safeness, Comfort, and Aesthetics criteria) influencing the walking experience, such as street lighting and road quality, as they will be explored in further detailed in the future.

5.2. Research Outlooks

The application-oriented experiment conducted in this research is a significant step in the process of transforming a conceptual urban walkability assessment tool into a real stakeholder-serving app. Considering the potential of software development, and the increase in the scale of GIS data, the presented method highlights how a GIS-based data processing program could be developed in the future that could provide useful information and analyses for defining choices oriented toward urban health. This application method of the MWM-Tool is an example of the transformation of a theoretical framework to a to practical one through a repeatable and scalable process that could be applied in other cities if sufficient urban GIS data are present (which is becoming more common). The described GIS-based approach is a faster and more objective way to collect data; additionally, the related maps are graphical resources that are able to provide immediate results. This GIS-based method is coherent, with a useful process of digitalization that facilitates understanding of urban phenomena, even for non-technical figures (Policy and Decision Makers). This points to the possibility of the systematic management of urban spatial resources, aimed at evaluating aspects capable of promoting the health and well-being of citizens, in a One Health perspective.

The health conditions to be understood are referred to as physical, social, and psychological well-being and are therefore related to multiple aspects within an urban context, i.e., not only to walkability. The next steps in our scientific research will be linked to a broader and more systemic evaluation of all aspects that contribute to health, such as the vulnerability of cities to the effects of climate change. Finally, efforts to connect urban data with public health data are under development. This work will make it possible to systemize geographical and morphological features with social (and health) related info. The proposed GIS-based method with the MWM-Tool could be a part of these efforts, being applied in each city in order to effectively guide strategic planning and programming choices.

To conclude, the MWM-Tool is a valuable tool that could be applied to different cities, especially those in Europe that have an urban environment similar to that of Milan. The author has already established a scientific network with colleagues and experts from Spain (Barcelona), France (Paris) and Germany (Berlin—where the framework was applied—and Hamburg) to increase the number of applications that will give substance to the methodology. The authors are conscious that the next challenge is to consider the power of Artificial Intelligence to enhance the data collection process to yield better predictions of urban mobility trends in the future.