Spatial pattern analysis of the impact of community food environments on foetal macrosomia, preterm births and low birth weight.

Community food environments (CFEs) have a strong impact on child health and nutrition and this impact is currently negative in many areas. In the Republic of Argentina, there is a lack of research evaluating CFEs regionally and comprehensively by tools based on geographic information systems (GIS). This study aimed to characterize the spatial patterns of CFEs, through variables associated with its three dimensions (political, individual and environmental), and their association with the spatial distribution in urban localities in Argentina. CFEs were assessed in 657 localities with ≥5,000 inhabitants. Data on births and CFEs were obtained from nationally available open-source data and through remote sensing. The spatial distribution and presence of clusters were assessed using hotspot analysis, purely spatial analysis (SaTScan), Moran's Index, semivariograms and spatially restrained multivariate clustering. Clusters of low risk for LBW, macrosomia, and preterm births were observed in the central-east part of the country, while high-risk clusters identified in the North, Centre and South. In the central-eastern region, low-risk clusters were found coinciding with hotspots of public policy coverage, high night-time light, social security coverage and complete secondary education of the household head in areas with low risk for negative outcomes of the birth variables studied, with the opposite with regard to households with unsatisfied basic needs and predominant land use classes in peri-urban areas of crops and herbaceous cover. These results show that the exploration of spatial patterns of CFEs is a necessary preliminary step before developing explanatory models and generating novel findings valuable for decision-making.

Spatial patterns of intestinal parasite infections among children and adolescents in some indigenous communities in Argentina.

Argentina has a heterogeneous prevalence of infections by intestinal parasites (IPs), with the north in the endemic area, especially for soil-transmitted helminths (STHs). We analyzed the spatial patterns of these infections in the city of Tartagal, Salta province, by an observational, correlational, and cross-sectional study in children and adolescents aged 1 to 15 years from native communities. One fecal sample per individual was collected to detect IPs using various diagnostic techniques: Telemann sedimentation, Baermann culture, and Kato-Katz. Moran's global and local indices were applied together with SaTScan to assess the spatial distribution, with a focus on cluster detection. The extreme gradient boosting (XGBoost) machine-learning model was used to predict the presence of IPs and their transmission pathways. Based on the analysis of 572 fecal samples, a prevalence of 78.3% was found. The most frequent parasite was Giardia lamblia (30.9%). High- and low-risk clusters were observed for most species, distributed in an east-west direction and polarized in two large foci, one near the city of Tartagal and the other in the km 6 community. Spatial XGBoost models were obtained based on distances with a minimum median accuracy of 0.69. Different spatial patterns reflecting the mechanisms of transmission were noted. The distribution of the majority of the parasites studied was aligned in a westerly direction close to the city, but the STH presence was higher in the km 6 community, toward the east. The purely spatial analysis provides a different and complementary overview for the detection of vulnerable hotspots and strategic intervention. Machine-learning models based on spatial variables explain a large percentage of the variability of the IPs.

A geospatial analysis of cardiometabolic diseases and their risk factors considering environmental features in a midsized city in Argentina.

New approaches to the study of cardiometabolic disease (CMD) distribution include analysis of built environment (BE), with spatial tools as suitable instruments. We aimed to characterize the spatial dissemination of CMD and the associated risk factors considering the BE for people attending the Non-Invasive Cardiology Service of Hospital Nacional de Clinicas in Córdoba City, Argentina during the period 2015-2020. We carried out an observational, descriptive, cross-sectional study performing non-probabilistic convenience sampling. The final sample included 345 people of both sexes older than 35 years. The CMD data were collected from medical records and validated techniques and BE information was extracted from Landsat-8 satellite products. A geographic information system (GIS) was constructed to assess the distribution of CMD and its risk factors in the area. Out of the people sampled, 41% showed the full metabolic syndrome and 22.6% only type-2 diabetes mellitus (DM2), a cluster of which was evidenced in north-western Córdoba. The risk of DM2 showed an association with high values of the normalized difference vegetation index (NDVI) (OR= 0.81; 95% CI: - 0.30 to 1.66; p=0.05) and low normalized difference built index (NDBI) values that reduced the probability of occurrence of DM2 (OR= -1.39; 95% CI: -2.62 to -0.17; p=0.03). Considering that the results were found to be linked to the environmental indexes, the study of BE should include investigation of physical space as a fundamental part of the context in which people develop medically within society. The novel collection of satellite-generated information on BE proved efficient.

Feature importance: Opening a soil-transmitted helminth machine learning model via SHAP.

In the field of landscape epidemiology, the contribution of machine learning (ML) to modeling of epidemiological risk scenarios presents itself as a good alternative. This study aims to break with the "black box" paradigm that underlies the application of automatic learning techniques by using SHAP to determine the contribution of each variable in ML models applied to geospatial health, using the prevalence of hookworms, intestinal parasites, in Ethiopia, where they are widely distributed; the country bears the third-highest burden of hookworm in Sub-Saharan Africa. XGBoost software was used, a very popular ML model, to fit and analyze the data. The Python SHAP library was used to understand the importance in the trained model, of the variables for predictions. The description of the contribution of these variables on a particular prediction was obtained, using different types of plot methods. The results show that the ML models are superior to the classical statistical models; not only demonstrating similar results but also explaining, by using the SHAP package, the influence and interactions between the variables in the generated models. This analysis provides information to help understand the epidemiological problem presented and provides a tool for similar studies.