Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study
What is the relationship between mortality and satellite images as elucidated through the use of Convolutional Neural Networks?Background: Following a century of increase, life expectancy in the United States has stagnated and begun to decline in recent decades. Using satellite images and street vie...
Guardado en:
Autores principales: | , , , , , |
---|---|
Formato: | article |
Lenguaje: | EN |
Publicado: |
Frontiers Media S.A.
2021
|
Materias: | |
Acceso en línea: | https://doaj.org/article/ac3b8190298d4396a8d380a7b45e2237 |
Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
id |
oai:doaj.org-article:ac3b8190298d4396a8d380a7b45e2237 |
---|---|
record_format |
dspace |
spelling |
oai:doaj.org-article:ac3b8190298d4396a8d380a7b45e22372021-11-05T06:00:08ZUsing Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study2296-256510.3389/fpubh.2021.766707https://doaj.org/article/ac3b8190298d4396a8d380a7b45e22372021-11-01T00:00:00Zhttps://www.frontiersin.org/articles/10.3389/fpubh.2021.766707/fullhttps://doaj.org/toc/2296-2565What is the relationship between mortality and satellite images as elucidated through the use of Convolutional Neural Networks?Background: Following a century of increase, life expectancy in the United States has stagnated and begun to decline in recent decades. Using satellite images and street view images, prior work has demonstrated associations of the built environment with income, education, access to care, and health factors such as obesity. However, assessment of learned image feature relationships with variation in crude mortality rate across the United States has been lacking.Objective: We sought to investigate if county-level mortality rates in the U.S. could be predicted from satellite images.Methods: Satellite images of neighborhoods surrounding schools were extracted with the Google Static Maps application programming interface for 430 counties representing ~68.9% of the US population. A convolutional neural network was trained using crude mortality rates for each county in 2015 to predict mortality. Learned image features were interpreted using Shapley Additive Feature Explanations, clustered, and compared to mortality and its associated covariate predictors.Results: Predicted mortality from satellite images in a held-out test set of counties was strongly correlated to the true crude mortality rate (Pearson r = 0.72). Direct prediction of mortality using a deep learning model across a cross-section of 430 U.S. counties identified key features in the environment (e.g., sidewalks, driveways, and hiking trails) associated with lower mortality. Learned image features were clustered, and we identified 10 clusters that were associated with education, income, geographical region, race, and age.Conclusions: The application of deep learning techniques to remotely-sensed features of the built environment can serve as a useful predictor of mortality in the United States. Although we identified features that were largely associated with demographic information, future modeling approaches that directly identify image features associated with health-related outcomes have the potential to inform targeted public health interventions.Joshua J. LevyJoshua J. LevyJoshua J. LevyRebecca M. LebeauxRebecca M. LebeauxAnne G. HoenAnne G. HoenAnne G. HoenBrock C. ChristensenLouis J. VaickusTodd A. MacKenzieTodd A. MacKenzieTodd A. MacKenzieFrontiers Media S.A.articledeep learningsatellite imagesmortalityremote sensingpublic healthPublic aspects of medicineRA1-1270ENFrontiers in Public Health, Vol 9 (2021) |
institution |
DOAJ |
collection |
DOAJ |
language |
EN |
topic |
deep learning satellite images mortality remote sensing public health Public aspects of medicine RA1-1270 |
spellingShingle |
deep learning satellite images mortality remote sensing public health Public aspects of medicine RA1-1270 Joshua J. Levy Joshua J. Levy Joshua J. Levy Rebecca M. Lebeaux Rebecca M. Lebeaux Anne G. Hoen Anne G. Hoen Anne G. Hoen Brock C. Christensen Louis J. Vaickus Todd A. MacKenzie Todd A. MacKenzie Todd A. MacKenzie Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study |
description |
What is the relationship between mortality and satellite images as elucidated through the use of Convolutional Neural Networks?Background: Following a century of increase, life expectancy in the United States has stagnated and begun to decline in recent decades. Using satellite images and street view images, prior work has demonstrated associations of the built environment with income, education, access to care, and health factors such as obesity. However, assessment of learned image feature relationships with variation in crude mortality rate across the United States has been lacking.Objective: We sought to investigate if county-level mortality rates in the U.S. could be predicted from satellite images.Methods: Satellite images of neighborhoods surrounding schools were extracted with the Google Static Maps application programming interface for 430 counties representing ~68.9% of the US population. A convolutional neural network was trained using crude mortality rates for each county in 2015 to predict mortality. Learned image features were interpreted using Shapley Additive Feature Explanations, clustered, and compared to mortality and its associated covariate predictors.Results: Predicted mortality from satellite images in a held-out test set of counties was strongly correlated to the true crude mortality rate (Pearson r = 0.72). Direct prediction of mortality using a deep learning model across a cross-section of 430 U.S. counties identified key features in the environment (e.g., sidewalks, driveways, and hiking trails) associated with lower mortality. Learned image features were clustered, and we identified 10 clusters that were associated with education, income, geographical region, race, and age.Conclusions: The application of deep learning techniques to remotely-sensed features of the built environment can serve as a useful predictor of mortality in the United States. Although we identified features that were largely associated with demographic information, future modeling approaches that directly identify image features associated with health-related outcomes have the potential to inform targeted public health interventions. |
format |
article |
author |
Joshua J. Levy Joshua J. Levy Joshua J. Levy Rebecca M. Lebeaux Rebecca M. Lebeaux Anne G. Hoen Anne G. Hoen Anne G. Hoen Brock C. Christensen Louis J. Vaickus Todd A. MacKenzie Todd A. MacKenzie Todd A. MacKenzie |
author_facet |
Joshua J. Levy Joshua J. Levy Joshua J. Levy Rebecca M. Lebeaux Rebecca M. Lebeaux Anne G. Hoen Anne G. Hoen Anne G. Hoen Brock C. Christensen Louis J. Vaickus Todd A. MacKenzie Todd A. MacKenzie Todd A. MacKenzie |
author_sort |
Joshua J. Levy |
title |
Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study |
title_short |
Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study |
title_full |
Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study |
title_fullStr |
Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study |
title_full_unstemmed |
Using Satellite Images and Deep Learning to Identify Associations Between County-Level Mortality and Residential Neighborhood Features Proximal to Schools: A Cross-Sectional Study |
title_sort |
using satellite images and deep learning to identify associations between county-level mortality and residential neighborhood features proximal to schools: a cross-sectional study |
publisher |
Frontiers Media S.A. |
publishDate |
2021 |
url |
https://doaj.org/article/ac3b8190298d4396a8d380a7b45e2237 |
work_keys_str_mv |
AT joshuajlevy usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT joshuajlevy usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT joshuajlevy usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT rebeccamlebeaux usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT rebeccamlebeaux usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT anneghoen usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT anneghoen usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT anneghoen usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT brockcchristensen usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT louisjvaickus usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT toddamackenzie usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT toddamackenzie usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy AT toddamackenzie usingsatelliteimagesanddeeplearningtoidentifyassociationsbetweencountylevelmortalityandresidentialneighborhoodfeaturesproximaltoschoolsacrosssectionalstudy |
_version_ |
1718444490135437312 |