Predicting Under-five mortality across 21 Low and Middle-Income Countries using Deep Learning Methods

Objectives To explore the efficacy of Machine Learning (ML) techniques in predicting under-five mortality in LMICs and to identify significant predictors of under-five mortality (U5M). Design This is a cross-sectional, proof-of-concept study. Settings and participants We analysed data from the Demographic and Health Survey (DHS). The data was drawn from 21 Low-and-Middle Income Countries (LMICs) countries (N = 1,048,575). Eligible mothers in each household were asked information about their children and the reproductive care they received during the pregnancy. Primary and secondary outcome measures The primary outcome measure was under-five mortality; secondary outcome was comparing the efficacy of deep learning algorithms: Deep Neural Network (DNN); Convolution Neural Network (CNN); Hybrid CNN-DNN with Logistic Regression (LR) for the prediction of child survival. Results We found that duration of breast feeding, household wealth index and the level of maternal education are the most important predictors of under-five mortality. We found that deep learning techniques are superior to LR for the classification of child survival: LR sensitivity = 0.47, specificity = 0.53; DNN sensitivity = 0.69, specificity = 0.83; CNN sensitivity = 0.68, specificity = 0.83; CNN-DNN sensitivity = 0.71, specificity = 0.83. Conclusion Our findings provide an understanding of interventions that needs to be prioritized, in order to reduce levels of U5M in LMICs. It also demonstrates that deep learning models are more efficacious than a traditional analytical approach. Strengths and limitations of this study The models were tested using a very large data sample, drawn from over 1 million households. The survey utilised a cluster sampling approach and are representative of each country included. Socio-economic, political and cultural differences between the included countries may limit generalisability of the results. The cross-sectional design of the study means we can only infer association and not causality.