2017
DOI: 10.1590/1516-3180.2016.0309010217
|View full text |Cite
|
Sign up to set email alerts
|

Comparison of machine-learning algorithms to build a predictive model for detecting undiagnosed diabetes - ELSA-Brasil: accuracy study

Abstract: CONTEXT AND OBJECTIVE: Type 2 diabetes is a chronic disease associated with a wide range of serious health complications that have a major impact on overall health. The aims here were to develop and validate predictive models for detecting undiagnosed diabetes using data from the Longitudinal Study of Adult Health (ELSA-Brasil) and to compare the performance of different machine-learning algorithms in this task. DESIGN AND SETTING:Comparison of machine-learning algorithms to develop predictive models using dat… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1

Citation Types

1
28
0
7

Year Published

2019
2019
2022
2022

Publication Types

Select...
4
4

Relationship

0
8

Authors

Journals

citations
Cited by 58 publications
(36 citation statements)
references
References 33 publications
1
28
0
7
Order By: Relevance
“…In the last decade, by constructing predictive models, an attempt to identify the factors that are potentially associated with the development of diabetes through data mining techniques has been made with some promising results in predicting or even capturing diabetes at its early stage [4,[7][8][9][10][11][12]. Among these techniques, the decision tree technique was widely used in the medical field in making diagnostic approaches during clinical practice [4,11,[13][14][15].…”
Section: Introductionmentioning
confidence: 99%
“…In the last decade, by constructing predictive models, an attempt to identify the factors that are potentially associated with the development of diabetes through data mining techniques has been made with some promising results in predicting or even capturing diabetes at its early stage [4,[7][8][9][10][11][12]. Among these techniques, the decision tree technique was widely used in the medical field in making diagnostic approaches during clinical practice [4,11,[13][14][15].…”
Section: Introductionmentioning
confidence: 99%
“…No Brasil, a utilização desses algoritmos em saúde pública ainda é incipiente. Como exemplo, pode-se citar o estudo de Olivera et al 15 que desenvolveu modelos preditivos de diabetes não diagnosticada a partir de dados de 12.447 adultos entrevistados para o Estudo ELSA (Estudo Longitudinal da Saúde do Adulto), utilizando cinco algoritmos de machine learning (regressão logística, redes neurais, naive bayes, método dos K vizinhos mais próximos e random forest).…”
Section: Machine Learning Para Análises Preditivas Em Saúde: Exemplo unclassified
“…De modo geral, todos os algoritmos avaliados alcançaram AUC ROC superior a 0,70 nos dados de teste, e os melhores resultados foram observados para os modelos de redes neurais, regressão logística penalizada de lasso e regressão logística simples, respectivamente. Resultados semelhantes foram observados por Olivera et al 15 no ajuste de modelos preditivos de diabetes não diagnosticada em que os melhores resultados de performance foram observados para o modelo de redes neurais e de regressão logística, sem diferenças relevantes. Diferenças na performance de modelos preditivos podem ser atribuídas às características dos dados, como a definição da resposta de interesse e a disponibilidade de variáveis candidatas a predi-tores, e às técnicas e aos métodos utilizados para construir e avaliar os modelos, como o refinamento de possíveis valores para os hiperparâmetros dos algoritmos mais flexíveis, na etapa de treinamento, e à definição dos bancos de treinamento e teste.…”
Section: Discussão E Conclusõesunclassified
See 1 more Smart Citation
“…Complementarily, graph analysis is a powerful approach to facilitate the understanding of the relations between a set of variables from a network perspective [6]. Nevertheless, in Brazil, the employment of these algorithms in public health is still scarce, although a few studies have been performed [7,8].…”
Section: Introductionmentioning
confidence: 99%