Transfer Learning for Urban Landscape Clustering and Correlation with Health Indexes

. Within the EU-funded Pulse project, we are implementing a data analytic platform designed to provide public health decision makers with advanced approaches to jointly analyze maps and geospatial information with health care data and air pollution measurements. In this paper we describe a component of such platform, designed to couple deep learning analysis of geospatial images of cities and some healthcare and behavioral indexes collected by the 500 cities US project, showing that, in New York City, urban landscape signi ﬁ cantly correlates with the access to healthcare services.


Introduction
Recent advances in machine learning and deep learning enable the design and implementation of novel data analysis pipelines that allow fusing heterogeneous data sources to extract novel insights and predictive patterns. These approaches seem particularly suitable to help increasing our insights in the relationships between the urban landscape of cities and the behavior of their residents, with particular focus on wellbeing and healthcare indexes. In this context, it can be of interest of health care planners and city decision-makers to have instruments able to find clusters of city areas that share similar urban structures and to analyze some behavioral indexes of their residents, in particular to see potential correlations and to plan similar interventions in the different clusters, even if such clusters contain areas that are geographically far. We have applied such approach in the context of the PULSE (Participatory Urban Living for Sustainable Environment) EU-project 1 . PULSE aims at developing a set of models and technologies to predict and manage public health problems in cities and promote health. It follows a participatory approach where citizen provide data through personal devices that are integrated with information from heterogeneous sources: open city data, health systems, urban sensors and satellites. The project deals with various issues concerning air quality, lifestyle and personal behavior and it aims to investigate the correlations between the exposure to atmospheric pollutants, the citizen habits and the health of the citizen themselves, focusing on asthma and type 2 diabetes. PULSE is being implemented in 5 major cities all over the world. Within PULSE, we are implementing a data analytic platform that will provide public health decision makers with advanced approaches to jointly analyze maps and geospatial information with health care data and air pollution measurements.
In this paper we will describe the results obtained with a prototypical component of such platform, designed to couple deep learning analysis of geospatial images of cities and some healthcare and behavioral indexes, showing that in New York City urban landscape significantly correlates with the access to healthcare services.

Deep Learning and Transfer Learning Models
Deep neural models provide flexible instruments to perform non-linear approximation of a variety of multivariate functions and to extract latent variables from a data set. In a nutshell, deep neural models are neural networks with many layers, able to map nonlinear functions with a number of parameters that is typically lower than their equivalent models with one layer only. Such models are particularly attractive since they can be used to perform clustering, regression and classification starting from data sets made of images, texts, time series.
In dependence of the nature of the input data set, different architectures can be exploited, ranging from the combination of many Convolutional layers in the case of images to the use of Long-term/Short-term networks in the case of time series and speech/text data.
Recently, an increasing number of papers are using deep learning to examine the relationships between the urban landscape and some environmental or citizens' behavioral data [1][2][3].
One of the main limits of deep learning models is related to the need of very large data sets in order to be able to gain advantage of their capability of encoding even the finest details that can be important to map input data, without getting trapped into noise and poor parameters estimates.
Rather interestingly, in order to deal with this problem, it is possible to resort to an increasing set of pre-trained deep learning models that can be used for the task of transfer learning [4], i.e. models that are able to represent the input space into a set of latent variables on the basis of a mapping mechanism, usually a deep neural network, learned on a large (external) data set, so that the relationships between such latent variables and the outcomes can be later learned on a specific and smaller data set. A well-known example is Inception-v3, a convolutional neural network trained on more than a million images from the ImageNet database (http://www.image-net.org). The network has 48 layers and can classify images into one thousand object categories, including trees and many animals. Another interesting example is represented by the Painters [5] networks, developed to automatically classify Paintings of famous artists. In principle, any of those methods can be used following the transfer learning paradigm to represent images coming from urban landscape of New York City.

Data and Methods
Our analysis is based on two data sources: NYC high resolution images and healthcare data coming from the 500 cities project [6]. NYC images have been collected by the "The National Agriculture Imagery Program" (NAIP) that acquires aerial imagery during the agricultural growing seasons in the continental United States. In particular, we have downloaded an image having an original resolution of 0.5 m and have downsampled it to 2 m which allows to have a fine-grained representation of the aerial urban landscape (see Fig. 1).
As it will be explained in the following, the reason for the downsampling is that the big image has been subdivided into tiles and the neural network adopted can accept images having maximum size of 299 pixel; we had to tune the ground resolution in order to have meaningful tiles, embracing a sufficiently-sized area. Health care data have been extracted from the repository made available by the 500 Cities project. "500 cities" is a collaboration between CDC, the Robert Wood Johnson Foundation, and the CDC Foundation 2 . The project provides city-and census tractlevel small area estimates for chronic disease risk factors (unhealthy behaviors), health outcomes, and clinical preventive service use for the largest 500 cities in the United States. NYC is divided in 2166 census tracts and the latest data available concerns 2017. The 27 chronic diseases measures provided by the project are listed in Table 1.
The measures include major risk behaviors that lead to illness, suffering, and early death related to chronic diseases and conditions, as well as the conditions and diseases that are the most common, costly, and preventable of all health problems.

The Data Analysis Pipeline
The pipeline implemented in our work is described in Fig. 2. The NAIP NYC image has been subdivided into image square blocks having size of 256 Â 256 pixels, corresponding to a 512 m edge. Therefore, it was possible to estimate the value of each of the 27 variables collected by "500 Cities" for each block. During this process, blocks out of the tracts or over the sea have been excluded, thus reducing the dataset. The images have been then processed by a pretrained deep model, thus extracting the final features for each image. Images are clustered by resorting to k-means clustering, and the clusters, confirmed with visual inspection, are associated to the healthcare indexes by statistical analysis.

Image Blocks
The NAIP NYC image has been subdivided into 8336 images of 256 Â 256 pixels. Each image is a square with edge equal to 512 m. It must be underlined that the original image is georeferenced, meaning that each single pixel is precisely located in space. The small derived tiles are georeferenced as well and can be effectively overlapped to the health and well-being maps. The images have been processed resorting to the Matlab Image Processing and Mapping toolboxes, which are capable of properly managing georeferenced images. Figure 3 shows some examples of the resulting images.
It is possible to note that some of the squares have white areas, corresponding to unmapped zones, due to the irregular borders of the image and to the presence of sea, rivers. Due to the availability of the vector map of the borders of NYC, we have been able to quantify, for each tile, the amount of its surface lying inside the borders of the city; we then filtered the original tile set and maintained only those having a minimal overlapping of 90%.

Estimation of the Healthcare Indexes for Each Image Block
The healthcare indexes of the 500 Cities database are collected for census tracts and NYC has, as already reported, 2166 census tracts. In order to carry out our analysis, Fig. 2. The data analysis pipeline we had to determine the value of the considered variables for each image block. In fact, a given tile overlaps, in general, several tracts. Therefore, we had to implement a simple estimator of the healthcare index of the block, as: where hci(j) is the value of the generic health care index for the j-th census tract and w j is the percentage of the image block covered by the mentioned tract. An example is shown in Figs. 3 and 4.
In order to properly quantify hcis, the blocks with a white area greater than 10% of the image have been removed. The final number of image blocks used for the following analysis has thus lowered to 2512 images. Each image has been then processed by resorting to a deep neural model to extract a set of latent features.

Deep Neural Networks Processing and Clustering
As a deep neural network model used for transfer learning, we have selected the network developed for the 2016 Painters by number competition [6]. In such competition the goal was to learn how to discriminate the authors of paintings between 1584 unique painters, starting from a training set of 79433 instances; the test set was composed of 23817 instances. In this case, a deep neural network model was learned, with 23 layers, mostly convolutional layers with some max pooling layer. The Painters network computes a layer of 2048 latent variables before the final discrimination layer implemented with a soft-max non-linear function. Those latent variables can be used as a way to embed generic images in the latent space. Therefore, using the software Orange (https://orange.biolab.si) and its Python pipeline, we have processed all image blocks with the Painters model, thus obtaining a final data matrix of 2512 examples with 2048 features.
Such features have been used to cluster the image blocks by resorting to the wellknown K-means clustering algorithm. The value of K has been derived with a grid search between 2 and 6 and taking the value that maximize the Silhouette coefficient.

Correlation and Statistical Analysis
The final step of the data analysis pipeline is represented by the search of statistical correlations between the clusters and their hcis. Univariate multinomial logistic regression was applied to estimate the probability to belong to a specific cluster given single variables' values. Multivariate multinomial logistic regression was performed after removal of samples characterized by missing values. A backward stepwise selection procedure based on AIC was applied to identify the most informative set of variables jointly modulating the probability to belong to the clusters. Multinomial logistic regression and the stepwise selection procedure were implemented in the R packages "nnet" and "stats", respectively. Analyses were performed by the R software tool version 3.5.1 (http://www.r-project.org).

Clustering
K-means was run on the 2512 instances with Euclidean distance and 10 reruns. 4 clusters were found to maximize Silhouette coefficient. The output of the clustering algorithm has been validated by analyzing the cluster distribution with the tSNE twodimensional mapping, as reported in Fig. 5. It is easy to see that the four clusters are in general well separated in the two-dimensional space 3 .
Thanks to visual inspection, it is possible to highlight that the four clusters well correspond to different urban landscapes. Cluster C1 corresponds to green areas, Cluster C2 to residential areas with small houses, Cluster C3 to industrial areas and larger buildings, Cluster C4 to residential with larger buildings. Four examples are shown in Fig. 6. Cluster analysis clearly show that the deep neural network model is able to map images in the latent space that share the intuitive notion of similarity that humans may use when they have to classify urban landscape. The method is thus able to automatically cluster similar areas where similar interventions can be planned.

Correlation and Statistical Analysis
Univariate analysis shows that 22 over 27 variables were significantly correlated with the clusters. This is also confirmed with visual inspection showing the variables distributions after equal frequency discretization against the clusters, as shown in Fig. 7.
In general, cluster C1, which is the one that groups green areas, has consistently better prevention and health indicators, but worse sleeping indexes and leisure time.
Overall, there is a gradient with all indexes moving from cluster C1, to C2, to C3 and finally to C4, which are the residential areas with large buildings.
A multivariate multinomial logistic regression has been performed to assess if significant correlations are present even in the multivariate setting. In this case, after a stepwise feature selection process, 20 variables have been selected. Of those, five variables have been found to be significant (p ( 0.01) in all sub-regressions performed by the multinomial model: Colon screening (Fecal occult blood test, sigmoidoscopy, or colonoscopy among adults aged 50-75 years), Chronic obstructive pulmonary disease among adults aged ! 18 years, High cholesterol among adults aged ! 18 years who have been screened in the past 5 years, Chronic kidney disease among adults aged ! 18 years and finally Stroke among adults aged ! 18 years.

Mapping
The blocks and the clusters have been represented in the original map, confirming the qualitative evaluation of the clusters reported above (Fig. 8). We can find green areas (C1), residential areas with larger buildings (C2), industrial areas (C3) and finally residential areas (C4).

Discussion and Conclusions
The data analysis pipeline described in this paper shows that it is possible to automatically correlate urban landscape with healthcare indicators at the whole city level. In the NYC case, such correlation seems particularly strong, probably because of social factors, which, in the US society, makes health indicators related to the urban areas where people live.
Our work has a number of implications.
First of all, it shows that deep neural networks designed to encode image data can be successfully reused within transfer learning approaches. Their application to represent urban landscape seems very effective.
Second, in the context of the PULSE project, the capability of finding clusters of similar urban landscape may allow to profile city areas, in which health care decision makers may plan similar interventions.
Finally, the combination of urban landscape and healthcare indicators is not only useful to hypothesize the intertwining of these two dimensions, but also to further profile urban areas by finding similar areas with similar behaviors of their inhabitants, thus allowing also life style interventions and more "precise" health care policies. Of course, the analysis has some limitations. First of all, the "quantification" of the health care indexes in the city blocks have been performed by a weighted averaging of the indexes of the census tracts included in the blocks. The weights are computed taking into account only the spatial overlap and not the actual number of inhabitants of the blocks. Second, the results obtained are probably "proxies" of the wealth of the people living in the different areas. For this reason, results may be representative of specific cities and not generalizable to other ones.