The most widely used measure is child mortality, for the health of children and an index of the general development of any country. Worldwide, under-five mortality rates are higher in South-Asian and Sub-Saharan African countries. The under-five mortality means is the probability of dying children before the fifth birthday. In India, the under-five child mortality rate is reduced by 49%, from 83 deaths per 1000 live births in 2000 to 42 deaths in 2017 [1].
The state-wide reports have been found that under-five mortality is highest in Uttar Pradesh, followed by Madhya Pradesh and Chhattisgarh [2], as shown in Figure 1.
Although there has been a significant reduction in these state deaths of under-five children, it remains a major issue for child health. However, completing the Sustainable Goal-3 is still a challenging task for developing countries like India. The death rate of an under-five child can be prevented with early predictive analytics like machine learning techniques. Nowadays, machine learning techniques are highly demanded in public health research. The various machine learning models have been used for prediction and classification methods in different types of health and biomedical data. The machine learning approaches could help to obtain early prediction and insight into the important factors for under-five mortality. These all-ML models can automatically identify interactions and can find the nonlinear relationship between dependent and independent variables. There are various machine learning prediction and classification models like regression, logistic regression, principal component analysis, decision tree, and maximum likelihood method that have been used to find the accurate estimation of causes of child mortality.
Previous studies have used the traditional models for predicting the under-five mortality factors. The study suggests that a weak health system is one of the major problems of maximum under-five death of children in low middle-income countries [3].
The research study used the Cox proportional hazard (CPH) model and frailty models to find under-five mortality risk factors [4]. The study has been applied classical and Bayesian approaches to the CPH model on under-five mortality data [5]. Another study has been done for finding the risk factors of under-five mortality with the help of the survival analysis method [6]. Within high-dimensional exposure data, machine learning (ML) approaches can be utilized to discover the exposures related to health outcomes of interest, as well as the potential interactions between those exposures [7]. The authors have been used the J48 and artificial neural network (ANN) techniques to find the causes of child mortality in Ethiopia [8]. The study was focused on assessing ML techniques' performance to predict the risk of neonatal mortality [9].
In the study, ML methods were used to predict neonatal morbidity and mortality [10]. ID3(Iterative Dichotomiser 3), random forest, and decision tree have been used for predicting the nutrition status in under-five age of children [11]. Another study was conducted in India to predict the nutrient effects on human health using ML techniques [12]. The research paper has been used a machine learning model to predict pre-term birth [13]. The above-cited research papers indicate that machine learning techniques have not been used in under-five mortality data. No single study is used for machine learning classification comparison in Indian and North Indian circumstances. Moreover, past studies have found a lack of a generic prediction framework for accurately estimating child mortality rates using machine learning techniques. There is a need for accurate prediction and classification models to provide highly accurate results and allow health researchers to experiment with various sets of aspects.
This study offers an opportunity to understand important factors and assess the accuracy of the machine learning techniques in under-five mortality. Apart from that, machine learning techniques explore the importance of predicting under-five mortality so that timely interventions can be made and factors that cause high rates of under-five mortality can be reduced.