Studying infant mortality rate: a data mining approach

Abstract

The World Health Organization (WHO) uses a large number of health predicators to measure a country’s overall health status. Among many, Infant Mortality Rate (IMR) is considered to be one important. Current literature describes the importance of these predictors in a much segregated manner. Also, its interrelationships are underexplored. This paper describes a framework for mining the hidden relationships of some predictors, such as (i) Population Annual Growth Rate (PAG), (ii) Total Fertility Rate (TFR), (iii) General Government Expenditure (GGE), (iv) Social Security Expenditure (SSE), (v) Total Expenditure on Health (THE), (vi) Access of Drinking Water (ADW), and (vii) Access to Basic Sanitation (ABS) with IMR with the help of Quantitative Association Rule (QAR) mining technique. Public database of WHO has been accessed to gather the required data region and country-wise and then normalized [0, 1]. Association and correlation rules considering the pair-wise influences of the predictors on IMR and vice versa have been engineered. The quality of each association has been measured using the concept of ‘leverage’. While, the term ‘lift’ interprets a correlation. The study finally reveals that rules showing association correlations of PAG, GGE, TFR and ADW positively predict for IMR status.