District-Level Food Environment Indicators and Social Vulnerability in São Paulo
Authors: Pedro Lemes Sixel Lobo, Eric Tokuda, Kuruvilla Joseph Abraham, Roberto Fray, Dirce Maria Marchioni, Alexandre Cláudio Botazzo Delbem, Rogerio Salvini
Organizations: Instituto de Informática, Universidade Federal de Goiás, Goiânia, GO, Brazil · Instituto de Ciências Matemáticas e de Computação, Universidade de São Paulo, São Carlos, SP, Brazil · Escola Superior de Agricultura Luiz de Queiroz da Universidade de São Paulo, USP, Piracicaba, SP, Brazil · Faculdade de Saúde Pública, USP, São Paulo, SP, Brazil
Urban food environments may reflect broader socioeconomic inequalities, but district-level evidence remains limited in Brazilian cities. This study examined whether indicators of food retail and street-market availability discriminate between levels of social vulnerability across the 96 districts of São Paulo. We conducted an exploratory cross-sectional ecological analysis integrating the São Paulo Social Vulnerability Index (IPVS), establishment records from the Relação Anual de Informações Sociais (RAIS), and street-market data from CAISAN. Census-sector information was aggregated at the district level. Twenty districts without an IPVS classification were excluded, resulting in 76 observations. The outcome distinguished districts classified as IPVS level 1 from those classified as levels 2--7. Predictors described the densities of healthy and unhealthy food establishments, the number of street markets, and the availability of establishments selling fresh or in natura food. Eight conventional machine-learning classifiers were evaluated using leave-one-out cross-validation. Reported mean F-scores ranged from 0.62 to 0.75, with XGBoost obtaining the highest value. In the Random Forest model, the densities of healthy and unhealthy food establishments jointly accounted for approximately 60% of the total impurity-based feature importance. These findings indicate that publicly available food-environment indicators contain information associated with the district-level distribution of social vulnerability. However, the small ecological sample, class imbalance, outcome binarization, and cross-sectional design limit predictive generalization and preclude causal or household-level interpretations.
Figures & tables
Compression
Comparison
Statistic
Significance
min
Mean–median
0.9219
0.00029997
min
Mean–no reduction
0.7770
9.999×10−5
min
Median–no reduction
0.9114
9.999×10−5
gzip
Mean–median
0.6425
0.00019998
gzip
Mean–no reduction
0.4476
0.0065993
gzip
Median–no reduction
0.8404
9.999×10−5
TABLE I: Mantel-test results comparing distance matrices obtained before and after district-level aggregation.
Index
IPVS category
# districts
0
Not classified
20
1
Extremely low vulnerability
56
2
Very low vulnerability
7
3
Low vulnerability
8
4
Medium vulnerability
1
5
High vulnerability (urban)
0
TABLE II: Distribution of IPVS categories across the 96 districts of São Paulo.
Variable
Type
Observed range
Density of healthy establishments
Continuous
8.31–192.41
Density of unhealthy establishments
Continuous
0.00–172.92
Number of street markets
Integer
0–22
Number of natural-food establishments
Integer
1–125
Establishments selling in natura food
Integer
3–147
São Paulo Social Vulnerability Index
Integer
0–7
TABLE III: Food-environment predictors and outcome used in the analysis.
Model
F-score
KNN
0.62 (0.49)
LDA
0.72 (0.45)
NB
0.71 (0.45)
DT
0.68 (0.46)
RF
0.70 (0.46)
GB
0.68 (0.46)
TABLE IV: Mean F-score and standard deviation obtained by the classification models under leave-one-out cross-validation.
Fig. 1: Impurity-based importance of the food-environment variables in the Random Forest model.
Model
Hyperparameters
KNN
leaf_size =1; n_neighbors =1; p =2; weights = uniform
Urban planning uses geospatial data, machine learning, and optimization to analyze cities and guide transport, land use, housing, and climate decisions.
Center for Professionalism and Value in Health Care, American Board of Family Medicine 1016 16th St NW Ste 800 Washington, DC 20036 United States of America · Stanford Center on Longevity, Stanford University Palo Alto, CA USA · Center for Population Health Sciences, School of Medicine, Stanford University Palo Alto, California, United States of America +1
Aug 8, 2026·Adrienne C. Kinney, Anya Workman, Ademar Takeo Akabane +5
Hastings Initiative for AI and Humanity, Bowdoin College, Brunswick, Maine, United States · Department of Digital and Computational Studies, Bowdoin College, Brunswick, Maine, United States · Postgraduate Program in Urban Infrastructure Systems, Polytechnique School, Pontifical Catholic University of Campinas, Campinas, São Paulo, Brazil +3
Operations Research Department, North Carolina State University, Raleigh, 27607, United States of America · Industrial and Systems Engineering Department, North Carolina State University, Raleigh, 27607, United States of America