INVESTIGATIVE MODELLING OF ANAEMIA IN PREGNANCY IN GHANA: A CASE STUDY IN BOLGATANGA

Abstract

Anaemia is the condition of low levels of haemoglobin in the blood, which results in a reduced amount of oxygen being transported in the body. In this study data on 580 pregnant women were obtained from the ante-natal care unit of the Bolgatanga regional hospital was used in investigating the presence of anaemia in pregnancy in the municipality and some of the factors which contribute significantly to anaemia in pregnancy using the Artificial Neural Network model, the Logit model and the Probit model to investigate and get best appropriate model for determining some of the factors and prevalence of anaemia in pregnant women and in the municipality respectively. The results revealed that the Artificial Neural Network model was the best model for investigating and determining some of the factors and prevalence of anaemia in pregnancy. This Artificial Neural Network model had the least A[C and SIC of 147.487 and 173.665 respectively across the Logit and Probit models, with a better predictive potential of 64.8%. The training and testing of the data for the Artificial Neural Network also showed the model being adequate. The Artificial Neural Network was proposed to be the best model for modelling anaemia in pregnancy. However, each model had it's limitation as there is no standard method for finding the ideal number of hidden nodes in a feed-forward network or for determining the best activation function for model equations when using the Artificial Neural Network. The Probit and Logit model equations limit the researcher by requiring a large amount of sample data to estimate the parameters of the equation and to find the data trend. Therefore more research needs to be done to get a better understanding of how both models are related.