Monitoring the evolution and di usion of diseases is an important task for health monitoring. Recent phenomena, such as the spread of Coronavirus (Covid-19), highlighted the relevance of adopting Geographical Information Systems (GIS) in epidemiology modeling. GIS may o er general views and indications which domain experts and governments could use to give immediate directions to social actors and operators. We report on the possibility of using geographic database model instruments in order to acquire, store and manage health-related data. The reported case study is about an application able to correlate TSH (Thyroid-Stimulating Hormone) with environmental data. The reported example aims to show how to acquire, analyze and integrate clinical and geographical data to evaluate possible correlations for the prevention of chronic diseases, especially neoplasms, by means of mapping disease features with environmental factors.
In the last years, GIS tools have become more useful in the public health sector,
especially in the epidemiology eld [1{3]. Epidemiology models represent an
important tool to manage and assess environmental risk factors, such as potential
cause of illness, by quantifying their impact on human health in population at
risk [
GIS can also be useful in associating epidemiological and health data by de ning how environmental factors a ect the health status of group of individuals or entire populations living in a territory, in the onset of speci c diseases [5{8, 0 Copyright c 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). This volume is published and copyrighted by its editors. SEBD 2020, June 21-24, 2020, Villasimius, Italy. ? Discussion Paper, a preliminary version of this paper has been published at IEEE
BIBM 2019 supported by POR PIH-GIS Regional Funds.
22]. GIS systems and analysis methodologies support domain experts in studying
the overlap of clinical and environmental data and can help with the identi
cation of possible relations between clinical features and diseases. These relations
can be rendered with rich epidemiological maps which represent the present and
future status (prediction) of the on-set of a disease and its human-to-human
transmission. Correlations between environmental risk factors and clinical data
or events can then be used to evaluate or foresee the impact on human health
and are usually be analyzed by using spatial analysis techniques [9{11]. These
techniques allow to study disease pathogenesis and etiology, identifying any links
between them and the places where they develop [
In literature, many contributions exist which focus on spatial
epidemiology [13{15]. In [
We present an example of GIS application to study and evaluate the
correlation between geographical and clinical geographical by using TSH
(ThyroidStimulating Hormone) data. Shape les of environmental layers have been used
in GIS to identify speci c territorial aspects and their in uence on clinical
outcomes. A dataset has been used containing data of patients enrolled at Magna
Graecia University of Catanzaro. A preliminary version of the paper has been
previously published in [
Environmental monitoring and the analysis of clinical data aim to prevent chronic
diseases, especially the neoplasms and their correlation with environmental
factors. In literature there exist many de nitions of both correlation (e.g. Pearson
correlation) and spatial correlation (scorr spatial correlation function between
two variables over an X-Y domain). For the purposes of the present work, we
adopt a loose de nition of spatial correlation as follows. Let's consider a clinical
event E in the input clinical dataset and a geographical entity G; we say that
E and G are spatially correlated if there exists a spatial query Q for which the
location of E is contained in the area subtended by G. In order to investigate on
spatial correlations in our dataset, di erent environmental layers have been used.
These are available on the Geoportal of the Calabria Region in shape le format
[
The online Geoportal of Calabria Region shares a cartographic data collection containing a number of environmental layers with IODL 2.0 license. Among the available layers, we selected the ones listed below as the most useful for our investigation: { urban area; { extra-urban area; { commercial and industrial settlements; { public and private service companies; { rural settlements; { mining area; { yards; { land ll; { scarp yards; { local railways; { airport areas and heliports; { regional, provincial and district administrative range.
These layers have been used to verify and analyze possible correlations between clinical and territorial aspects. The application of shape les and the use of geographic layers are important to assess whether and how some pathologies or in any case clinical alterations of individuals can be directly or indirectly related to strictly territorial aspects. As case study, we report the study of the correlation between the patterns of topological data with an example of thyroid's data regarding a south Italian region patients. A Health-GIS system has been implemented aiming to cross-check the TSH value with further environmental data through a topological overlay operation. The goal is to identify the geographical areas with the greatest presence of cases with speci c pathology. Moreover, it is also interesting to calculate and trace the average distance covered by a patient for a speci c health service. 3
The health dataset regards single le with .xls extension in which each record represents a patient and contains: { geo-referenced data (district, home address, latitude, longitude and altitude) { clinical data (TSH level expressed in U=ml)
The clinical dataset contains 2269 records and Table 1 reports an extract of the database we used for the investigation. The QGIS platform has been used to cross-check these clinical data with data relating to environmental factors. QGIS is a free and open source geographic information system to visualize and overlap vector and raster data in di erent formats and projections without any conversion into a common internal format. The clinical dataset has been converted in .cvs le and loaded into the QGIS system to represent spatial and clinical information together in a geographical map. Patients locations and other logistical data are geo-referenced and related with clinical data. For instance, red dots in Figure 1 refer to patients' location data extracted from the considered clinical dataset and projected on extractive areas polygons.
Starting from this representation, environmental layers listed above have been correlated with clinical dataset and some of these correlations are reported below.
Since several diseases have a close link with overexposure to toxic substances, it is relevant to measure the occurrence of patients locations with potential toxic areas. Air pollution is one of the major environmental risks and in 2015 it has been recognized by the World Health Assembly (WHA) as one of the world's major public health problems. Pollutants are elements and chemical compounds deriving mostly from human activities such as, for example, industrial and commercial production. For this reason, the onset of several cancer cases in a speci c geographic area has led to the urge to investigate and monitor industrial activities insisting on the same area.
Figure 1 shows the topological overlay between clinical data (points in red) and extractive areas (dark green areas) also reporting urban areas (areas in yellow). The extraction areas generate a series of pollutants whose exposure has been associated with a series of harmful e ects on human health, especially for the respiratory and cardiovascular system and for neoplasms. Pollutants are mainly responsible for a substantial impoverishment of the soil (e.g., caused by deforestation and intensive agriculture), a continuous deterioration of water quality (e.g., caused especially by industrial discharges), and a growing contamination of the air (e.g., caused from the presence of ne dust and particulate in smog). Many substances such as, for example, nitrogen oxide, sulfur oxide, ozone and dust act as irritants; pollen and other agents present in the air are responsible for allergies such as asthma and seasonal colds; some toxic agents such as benzene and polycyclic aromatic hydrocarbons are carcinogenic; carbon monoxide compromises the transport of oxygen by the blood with serious e ects on the brain. Other metals come into contact with various organs and tissues causing biological alterations and causing damage to the heart and the apparatus.
Figure 2 shows the topological overlay of clinical data with the level of landlls and scrap deposits (areas in brown), always bringing the urban area (areas in yellow) as reference. The economic growth of the most industrialized countries has generated a substantial increase in consumption and urbanization with a consequent increasing in waste production and in its di cult regular disposal. Waste has negative consequences on the environment and on human health. Asbestos is harmful to human health; breathing dust containing asbestos bers can cause serious diseases, pleural tumors (pleural mesothelioma) and lung cancer. 4
Spatial information in clinical data is crucial for prevention and monitoring evolution of global epidemiology. In this paper we analyze environmental data related with health related ones. We are working to complete a model and general purpose system able to relate clinical, environmental and epidemiology related information for real time data analysis and simulation. The aim is to support prevention in a large scale context. Our framework is general enough to support the loading and preprocessing of generic datasets in standard format (e.g. csv). We are planning to integrate user-friendly functionalities to import di erent datasets. We are also working on consolidating the representation of time as an analysis dimension and re ning its use in the data analysis pipeline by adopting time series analysis techniques.
This research has been supported by POR CALABRIA FESR-FSE 2014-2020 PIH-GIS project CUP J88C17000320006. The authors thank Giuseppe Brescia for his contribution to a preliminary version of the paper.