Scholarlink

Background <p>Each year, approximately 2.5 million newborns die globally, with developing countries bearing the impact of this crisis. Sub-Saharan Africa has the highest neonatal mortality rate, with Ethiopia facing alarmingly high figures, particularly in rural areas where mortality is significantly higher due to poor healthcare access and socio-economic challenges.</p> Methods <p>This study aimed to develop a predictive model for neonatal mortality in rural Ethiopia using secondary data from the Ethiopian Demographic and Health Surveys (2000–2019). The dataset included 29,048 instances and 22 relevant features, which were preprocessed to handle missing values and balance the class distribution using the Synthetic Minority oversampling technique. Several ensemble machine-learning algorithms, including Random Forest, Gradient Boosting, Extreme Gradient Boosting, Light Gradient Boosting, and CatBoost, were applied to build the model. Additionally, the logistic regression algorithm was employed to enhance transparency and interpretability and for comparative analysis. Model performance was evaluated based on accuracy, precision, recall, F1 score, and Receiver Operating Characteristic—Area Under the Curve.</p> Results <p>Among the algorithms tested, categorical boosting achieved the highest performance with 97.5% accuracy, 97.52% precision, 97.5% recall, 97.5% F1 score, and an exceptional Receiver Operating Characteristic—Area Under the Curve value of 99.57%. Key risk factors identified include BCG vaccination status, the number of under-five children in the household, recent diarrhea episodes, and iron tablet intake during pregnancy. Valuable feedbacks from community health workers were provided on these factors, helping to refine their impact on neonatal mortality.</p> Conclusions <p>This study developed an effective predictive model for neonatal mortality in rural Ethiopia, providing actionable insights for targeted interventions. The model underscores the importance of improving healthcare access, maternal health, and policy reforms, with the potential to reduce neonatal mortality through mobile health apps and policymaker collaboration.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predicting neonatal mortality using ensemble machine learning algorithms in the case of Ethiopian Rural Areas

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年8月22日

查看全文
  • Melaku Alelign Mengstie,
  • Misganaw Telake Telele

摘要

Background

Each year, approximately 2.5 million newborns die globally, with developing countries bearing the impact of this crisis. Sub-Saharan Africa has the highest neonatal mortality rate, with Ethiopia facing alarmingly high figures, particularly in rural areas where mortality is significantly higher due to poor healthcare access and socio-economic challenges.

Methods

This study aimed to develop a predictive model for neonatal mortality in rural Ethiopia using secondary data from the Ethiopian Demographic and Health Surveys (2000–2019). The dataset included 29,048 instances and 22 relevant features, which were preprocessed to handle missing values and balance the class distribution using the Synthetic Minority oversampling technique. Several ensemble machine-learning algorithms, including Random Forest, Gradient Boosting, Extreme Gradient Boosting, Light Gradient Boosting, and CatBoost, were applied to build the model. Additionally, the logistic regression algorithm was employed to enhance transparency and interpretability and for comparative analysis. Model performance was evaluated based on accuracy, precision, recall, F1 score, and Receiver Operating Characteristic—Area Under the Curve.

Results

Among the algorithms tested, categorical boosting achieved the highest performance with 97.5% accuracy, 97.52% precision, 97.5% recall, 97.5% F1 score, and an exceptional Receiver Operating Characteristic—Area Under the Curve value of 99.57%. Key risk factors identified include BCG vaccination status, the number of under-five children in the household, recent diarrhea episodes, and iron tablet intake during pregnancy. Valuable feedbacks from community health workers were provided on these factors, helping to refine their impact on neonatal mortality.

Conclusions

This study developed an effective predictive model for neonatal mortality in rural Ethiopia, providing actionable insights for targeted interventions. The model underscores the importance of improving healthcare access, maternal health, and policy reforms, with the potential to reduce neonatal mortality through mobile health apps and policymaker collaboration.