Efficacy of deep learning methods for predicting under-five mortality in 34 low-income and middle-income countries

Abstract

ObjectivesTo explore the efficacy of machine learning (ML) techniques in predicting under-five mortality (U5M) in low-income and middle-income countries (LMICs) and to identify significant predictors of U5M.DesignThis is a cross-sectional, proof-of-concept study.Settings and participantsWe analysed data from the Demographic and Health Survey. The data were drawn from 34 LMICs, comprising a total of n=1 520 018 children drawn from 956 995 unique households.Primary and secondary outcome measuresThe primary outcome measure was U5M; 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’s survival.ResultsWe found that duration of breast feeding, number of antenatal visits, household wealth index, postnatal care and the level of maternal education are some of the most important predictors of U5M. 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.ConclusionOur findings provide an understanding of determinants of U5M in LMICs. It also demonstrates that deep learning models are more efficacious than traditional analytical approach.

Use of a deep learning and random forest approach to track changes in the predictive nature of socioeconomic drivers of under-5 mortality rates in sub-Saharan Africa

“…2 3 25 52-54 Ezeh et al, 54 found that mother's education level and household wealth influenced child survival in Nigeria. A similar study by Adegbosin et al, 25 that used deep learning techniques in predicting U5MR in low-inome and middle-income countries, ranked mother's education and household wealth index among the most critical predictors of U5MR. The same study found that deep learning techniques are superior in predicting child survival, and a similar conclusion has been arrived at in other similar studies.…”

“…The survey uses a two-stage cluster sampling. 25 More information about the sampling design, data collection and processing details are described on the DHS programme website. The datasets are available on request from the DHS programme.…”

“…[18][19][20][21][22][23][24] These machine learning models are known to be very flexible compared with the statistical models like the CPH model. [21][22][23][24][25] A recent study by Adegbosin et al, 25 recommended using deep learning models to understand the determinants of U5MR in low-income and middle-income countries.…”

“…Previous studies have shown that the four socioeconomic factors; place of residence, mother's education, household wealth index and sex of the child, are often stated among the top predictors of under-5 mortality in the sub-Saharan region. [12][13][14][15][16][17][18][19][20][21][22][23][24][25] With the launch of the MDG in the year 2000, we saw the convergence of the development agendas of United Nations Development Programme; United Nations Environment Programme; WHO; UNICEF; UNESCO and other development agencies, to raise funding and create programmes to combat existing inequalities to achieve these goals. 26 Despite the substantial improvement made with the MDG4, inequalities persist today, and progress has been uneven.…”

“…Recently, deep learning methods, which are advances in neural networks, have been recommended for analysing survival data 18–24. These machine learning models are known to be very flexible compared with the statistical models like the CPH model 21–25. A recent study by Adegbosin et al ,25 recommended using deep learning models to understand the determinants of U5MR in low-income and middle-income countries.…”