Using Undersampling with Ensemble Learning to Identify Factors Contributing to Preterm Birth

TL;DR

This paper develops ensemble learning models with novel missing data handling and undersampling techniques to identify key factors contributing to preterm birth, improving sensitivity significantly.

Contribution

It introduces new ensemble feature selection methods incorporating missing data and class imbalance handling, advancing analysis of preterm birth factors.

Findings

42% improvement in sensitivity over previous methods

Effective handling of missing data and class imbalance

Comparison of multiple ensemble feature selection techniques

Abstract

In this paper, we propose Ensemble Learning models to identify factors contributing to preterm birth. Our work leverages a rich dataset collected by a NIEHS P42 Center that is trying to identify the dominant factors responsible for the high rate of premature births in northern Puerto Rico. We investigate analytical models addressing two major challenges present in the dataset: 1) the significant amount of incomplete data in the dataset, and 2) class imbalance in the dataset. First, we leverage and compare two types of missing data imputation methods: 1) mean-based and 2) similarity-based, increasing the completeness of this dataset. Second, we propose a feature selection and evaluation model based on using undersampling with Ensemble Learning to address class imbalance present in the dataset. We leverage and compare multiple Ensemble Feature selection methods, including Complete Linear Aggregation (CLA), Weighted Mean Aggregation (WMA), Feature Occurrence Frequency (OFA), and Classification Accuracy Based Aggregation (CAA). To further address missing data present in each feature, we propose two novel methods: 1) Missing Data Rate and Accuracy Based Aggregation (MAA), and 2) Entropy and Accuracy Based Aggregation (EAA). Both proposed models balance the degree of data variance introduced by the missing data handling during the feature selection process while maintaining model performance. Our results show a 42\% improvement in sensitivity versus fallout over previous state-of-the-art methods.