Potential Confounders in the Analysis of Brazilian Adolescent’s Health: A Combination of Machine Learning and Graph Theory

Abstract

The prevalence of health problems during childhood and adolescence is high in developing countries such as Brazil. Social inequality, violence, and malnutrition have strong impact on youth health. To better understand these issues we propose to combine machine-learning methods and graph analysis to build predictive networks applied to the Brazilian National Student Health Survey (PenSE 2015) data, a large dataset that consists of questionnaires filled by the students. By using a combination of gradient boosting machines and centrality hub metric, it was possible to identify potential confounders to be considered when conducting association analyses among variables. The variables were ranked according to their hub centrality to predict the other variables from a directed weighted-graph perspective. The top five ranked confounder variables were "gender", "oral health care", "intended education level", and two variables associated with nutrition habits-"eat while watching TV" and "never eat fast-food". In conclusion, although causal effects cannot be inferred from the data, we believe that the proposed approach might be a useful tool to obtain novel insights on the association between variables and to identify general factors related to health conditions.

A scoping review on the use of machine learning in research on social determinants of health: Trends and research prospects

“…Regarding the origin of the data sources, more than half of the studies (n = 46, 56%) used US data, followed by global data (n = 9, 11%). Several studies were based on data from Australia (n = 5) ( Bentley et al, 2018 ; Cramb, Mengersen, & Baade, 2011 ; Handley et al, 2014 ; Hu et al, 2009 , 2010 ), the UK (n = 3) ( Engchuan et al, 2019 ; Penny and Smith, 2012 ; Suel et al, 2019 ), Iran (n = 2) ( Bastaminia, Rezaei, & Saraei, 2017 ; Darvishi et al, 2017 ), and Brazil (n = 2) ( Ambriola Oku et al, 2020 ; Chiavegatto Filho et al, 2018 ).…”

“… Prediction Survey Morocco 356 Bone mineral density CART a , CTA a A lower level of education was associated with significantly lower bone mineral densities at the lumbar spine and the hip sites and with a higher prevalence of osteoporosis at these sites in a dose-response manner. ( Ambriola Oku et al, 2020 ) 2019 To explore potential confounders in an adolescent public health dataset of a developing country by using a combination of machine learning methods and graph analysis. Prediction Survey Brazil 102301 Health status Gradient boosting machines The proposed approach might be a useful tool to obtain novel insights on the association between variables and to identify general factors related to health conditions.…”

Conectando grafos, grupos e termodinâmica via deformações no Laplaciano para análise de dados em redes complexas

“…Seja na determinação do desempenho socioeconômico de indivíduos ou na identificação de fatores de saúde que impactam conjuntos de doenças. 83 Para abordar esses novos desafios, estudos recentes [84][85][86][87] propõem abordagens de mapeamento de dados tabulares em grafos visando uma melhor análise exploratória dos dados. Em alguns estudos, como, 84,85,87 essas relações são expressas por arestas não direcionadas, cujos pesos são determinados com base em medidas de informação mútua ou valores de correlação.…”

“…83 Para abordar esses novos desafios, estudos recentes [84][85][86][87] propõem abordagens de mapeamento de dados tabulares em grafos visando uma melhor análise exploratória dos dados. Em alguns estudos, como, 84,85,87 essas relações são expressas por arestas não direcionadas, cujos pesos são determinados com base em medidas de informação mútua ou valores de correlação. No entanto, a construção desses grafos pode resultar em vários componentes desconexos, um problema se as relações entre todos os pares de vértices forem importantes.…”