sc 22(18): e1

Research Article

Predicting Diabetes Disease for healthy smart cities

Download11 downloads
  • @ARTICLE{10.4108/eetsc.v6i18.589,
        author={Hugo Peixoto and Vasco Ramos and  Carolina Marques and Jos\^{e} Machado},
        title={Predicting Diabetes Disease for healthy smart cities},
        journal={EAI Endorsed Transactions on Smart Cities},
        volume={6},
        number={18},
        publisher={EAI},
        journal_a={SC},
        year={2022},
        month={4},
        keywords={Data Mining, Diabetes, CRISP-DM, Classification, ML Models, Smart Cities, Smart Health},
        doi={10.4108/eetsc.v6i18.589}
    }
    
  • Hugo Peixoto
    Vasco Ramos
    Carolina Marques
    José Machado
    Year: 2022
    Predicting Diabetes Disease for healthy smart cities
    SC
    EAI
    DOI: 10.4108/eetsc.v6i18.589
Hugo Peixoto1,*, Vasco Ramos1, Carolina Marques1, José Machado1
  • 1: University of Minho
*Contact email: hpeixoto@di.uminho.pt

Abstract

INTRODUCTION: Diabetes is a chronic condition that affects a large portion of the population and is the leading cause of numerous health problems. Its automatic detection could improve the communities’ overall well-being. OBJECTIVES: The primary goal was to introduce advancements to the subject of healthy smart cities by studying an approach for predicting the occurrence of diabetes in the Pima Female Adult Population using data mining. METHODS: This study uses CRISP-DM to analyze the results of six different models acquired from three different iterations of the same dataset. DISCUSSION: This study found that the most promising model is k-NN, which obtained results of almost 92% of F1 Score with the third data preparation strategy. CONCLUSION: Acceptable results were achieved with the k-NN model and the third data preparation strategy, but more research into improving the data preparation processes and their influence on the outputs of each model is needed.