Abstract:
Introduction: Early Neonatal mortality refers to the death of a newborn less than seven days
of life. Despite significant reductions in recent years, neonatal mortality remains a public
health problem in middle and low-income countries. Early detection and treatment are critical
for avoiding complications that lead to neonatal death and increasing the quality of life.
The most challenging task in treating early neonatal death is determining the underlying cause
of the Death. This study aims to address this gap by implementing different machine-learning
classification methods to predict the cause of early neonatal mortality.
Objective: To predict early neonatal mortality and identify its predictors in East Africa
2025GC
Method: This study used secondary data analysis based on based on DHS data collected using
cross sectional study design in East African countries. from a recent demographic health
survey in an East African country. A weighted sample of 478,549 live births were included in
the study. Machine learning algorithms were used in the study to analyze 2016 to 2024
demographic and health surveys. Data was cleaned by STATA 17, using Python (version
3.12) to process the data, and the model was evaluated by area under the curve, accuracy,
precision, recall, and F-measure.
Results: Early neonatal mortality in early Africa was 22 deaths per 1,000 live births. The
Extra tree model outperformed with 96.7% accuracy and 98% area under the curve in
predicting early neonatal mortality. The Shapley additive explanation feature importance plot
of the optimized for extra tree model Maternal Anemia, inadequate ANC visit, inadequate
Vaccination TT Vaccination, no formal Maternal Education, and mobile were the top
predictors of Early neonatal mortality.
Conclusion: In this study, we assessed the ten machine learning algorithms for detecting
Early neonatal mortality. To reduce the prevalence of early neonatal mortality, this finding
recommends to emphasis on improving anemia prevention and control strategy services,
improving ANC services, improving health education, and awareness to reduce early neonatal
mortality
