Exploring the risk factors of preterm birth using data mining
Research highlights
► Multiple birth is the most important risk factor of preterm. ► Hemorrhage during pregnancy is a major predicted factor of preterm for single birth. ► Paternal smoking is a potential risk factor of preterm.
Introduction
Preterm birth, the birth of an infant prior to 37 completed weeks of gestation, is the leading cause of perinatal morbidity and mortality (Goldenberg et al., 2008, McCormick, 1985). The prevalence rate for such birth is about 12–13% in the USA, and 5–9% in Europe, other developed countries and Taiwan (Chuang et al., 2007, MacDorman et al., 2005, Slattery and Morrison, 2002). The reasons for preterm birth remain unclear, although data mining is a promising approach to explore potential factors from large amount of data (Chang, 2007, Chen et al., 2009, Courtney et al., 2008, Liao et al., 2008). Hence, the purpose of this work, based on the nest case-control study design, is to explore the risk factors of preterm by neural network and decision tree in data mining, to find more potential information.
Section snippets
Preterm birth
Preterm birth is the birth of an infant within 37 weeks of gestation, which accounts for 75% of perinatal mortality and half the long-term morbidity (McCormick, 1985). Studies show that maternal race, age, weight, income, previous preterm history, weight gain, infection, stress during pregnancy and other immunologically medicated processes are the risks factors for such birth (Goldenberg et al., 2008, Moore, 2003, Romero et al., 2006). However, despite the identification of such factors, a
Research structure
Clementine 10.0 is a commercial data mining tool (SPSS, 2005) that supports various analyses, such as neural network, decision tree, regression, logistic, and so on. To analyze the nominal variables, we used the neural network and C5.0 of Clementine 10.0 to mine the data.
The research procedures were as follow:
- 1.
Problem definition: Preterm birth is one of the leading causes of diseases and death among newborns. In addition, preterm infants often suffer long-term health problems, including lung
Research results
The relative importance of inputs, as derived by the neural network, is shown in Table 1. Because a lot of variables were measured in the current medical data, we decided to explore it in two stages. First, a neural network was used to investigate the risk factors of preterm birth, and the 15 top important factors, with coefficients larger than 0.0300, were than used in the next stage of our study. These factors were as follows: number of birth, paternal smoking, hemorrhage during pregnancy,
Conclusions
Preterm birth is the leading cause of perinatal morbidity and mortality, but to date the precise mechanism is still unknown now, and few prospective studies have been undertaken that explore the risk factors using data mining methods. Hence, we conducted a study based on the nest case-control design method and used data mining to explore risk factors of preterm. Our results show that multiple birth and hemorrhage during pregnancy are the top two risk factors. In addition, several maternal
Acknowledgements
We thank professor Jung-Der Wang, Pau-Chung Chen and Hui-I Hsieh for providing us medical health data and helpful comments.
References (28)
- et al.
Data mining approach for supply unbalance detection in induction motor
Expert Systems with Applications
(2009) A study of applying data mining to early intervention for developmentally-delayed children
Expert Systems with Applications
(2007)- et al.
Efficient sleep spindle detection algorithm with decision tree
Expert Systems with Applications
(2009) - et al.
Epidemiology and causes of preterm birth
Lancet
(2008) - et al.
A case study of applying data mining techniques in an outfitter’s customer value analysis
Expert Systems with Applications
(2009) - et al.
Mining product maps for new product development
Expert Systems with Applications
(2008) - et al.
Applying enhanced data mining approaches in predicting bank performance: A case of Taiwanese commercial banks
Expert Systems with Applications
(2009) - et al.
Artificial neural networks for optimization of gold-bearing slime smelting
Expert Systems with Applications
(2009) Preterm labor and birth: what have we learned in the past two decades?
Journal of Obstetric Gynecologic & Neonatal Nursing
(2003)- et al.
Automatic classification of Tamil documents using vector space model and artificial neural network
Expert Systems with Applications
(2009)
Preterm delivery
Lancet
An expert system to predict protein thermostability using decision tree
Expert Systems with Applications
Classification and regression trees
Cited by (48)
Data-Driven Modeling of Pregnancy-Related Complications
2021, Trends in Molecular MedicineIdentification of blood glucose patterns in patients with type 1 diabetes using continuous glucose monitoring and clustering techniques
2021, Endocrinologia, Diabetes y NutricionIntelligent system based on data mining techniques for prediction of preterm birth for women with cervical cerclage
2020, Computational Biology and ChemistryCitation Excerpt :They demonstrated that linear SVM provided a robust base line of performance with sensitivity rate of 40% and G-Means of 57–60%. Chen et al. investigated the risk factors that may lead to preterm birth by using two classifiers: neural networks and decision tree C5.0 (Chen et al., 2011). The dataset were collected prospectively for 910 mothers–child participants from National Taiwan University Hospital in Taiwan.
SKA2 gene – A novel biomarker for latent anxiety and preterm birth prediction
2019, European Journal of Obstetrics and Gynecology and Reproductive BiologyCitation Excerpt :The cortisol levels, the anxiety scores as well as the genotypes of AA and AG were significantly higher in the early PTB group than the late PTB group (P < 0.001). The causations of PTB remained elusive [23]. The parturition process is an uncontrollable and incurable condition because the parturition is a one-way process.
- 1
These authors contributed equally to the work.