← Back to all news
arXiv Machine Learning By Nathaniel Hendrix, Carl Y. Zhang, Chris Heitzig, Andrew Bazemore, David H. Rehkopf

Geospatial Foundation Models Capture Health-Relevant Dimensions of Place Beyond Conventional Social Risk Indices

Read the original on arXiv Machine Learning →

The study evaluates whether geospatial foundation models derived from 2022 satellite data can capture physical aspects of place that conventional area-based social risk indices miss. Using LightGBM on data from 82,646 census tracts, the models moderately predicted certain survey variables and explained up to 54% of the residual variance in 40 health outcomes, notably improving predictions for annual checkups, arthritis, and high blood pressure. The models’ explanatory power increased with larger tract sizes, suggesting they add valuable, health-relevant information beyond traditional social risk measures.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

Related stories

arXiv Machine Learning
Aug 28

District-Level Food Environment Indicators and Social Vulnerability in S\~ao Paulo

This study examined whether food retail and street‑market indicators differentiate social vulnerability levels across 76 districts of São Paulo. Using machine‑learning classifiers on district‑level data, the authors found that densities of healthy and unhealthy food establishments explained about 60 % of feature importance, with XGBoost achieving the highest mean F‑score (0.75). The results suggest that publicly available food‑environment data are linked to district‑level social vulnerability, though limitations such as small sample size and cross‑sectional design restrict causal inference.

By Pedro Lemes Sixel Lobo, Eric Tokuda, Kuruvilla Joseph Abraham, Roberto Fray, Dirce Maria Marchioni, Alexandre Cl\'audio Botazzo Delbem, Rogerio Salvini
More like this →
arXiv AI
Sep 15

Transfer Learning for Socioeconomic Estimation in Forced-Displacement Settings

The paper presents a transfer‑learning approach that adapts a multimodal spatiotemporal vision transformer, originally trained on Demographic and Health Survey data, to estimate socioeconomic conditions in forced‑displacement settings. Using satellite‑derived geospatial covariates, the adapted model explains up to 66% of variation in socioeconomic outcomes in camp‑intersecting grids and 41% in non‑camp areas, achieving mean absolute errors of 4.37 and 5.41 index points respectively. This framework supplements periodic household surveys by providing regularly updated, spatially granular socioeconomic estimates that bridge data gaps between survey rounds.

By Steven Ndung'u, Adel Daoud, Ismael Yacoubou Djima, Hai-Anh H. Dang, Patrick Michael Brock
llmscomputer-visionfine-tuningmultimodalbenchmarks
More like this →
arXiv Machine Learning
Jun 18

UST-GNN: A Unified Spatial--Topological Graph Neural Network Framework for Urban Analytics--Demonstrated through a Case Study on Urban Health Prediction

arXiv:2504. 04739v3 Announce Type: replace Abstract: Understanding how social, demographic, environmental, and spatial factors jointly shape urban outcomes is essential for sustainable urban development and evidence-based policy.

By Minwei Zhao, Sanja Scepanovic, Stephen Law, Ivica Obadic, Cai Wu, Daniele Quercia
ragsafety
More like this →