Comparison of Various Data Mining Algorithms in the Prediction of Risk for Gestational Diabetes

Abstract

Data Mining is a field of computer science which is used to discover new patterns for large data sets. Classification is an important task in data mining. In different areas of medicine, data mining has contributed to improve the results with other methodologies. Gestational diabetes is a condition characterized by high blood sugar (glucose) levels that is first recognized during pregnancy period of a woman. Diabetes is a disease in which levels of blood glucose, also called blood sugar, are above normal. People with diabetes have problems converting food to energy. Normally, after a meal, the body breaks food down into glucose, which the blood carries to cells throughout the body. Cells use insulin, a hormone made in the pancreas, to help them convert blood glucose into energy.

During the second and third trimester, a mother's diabetes can lead to over-nutrition and excess growth of the baby. Having a large baby increases risks during labour and delivery. For example, large babies often require caesarean deliveries and if he or she is delivered vaginally, they are at increased risk for trauma to their shoulder. In addition, when foetal over-nutrition occurs and hyper insulinemia results, the baby's blood sugar can drop very low after birth, since it won't be receiving the high blood sugar from the mother. However, with proper treatment, a gestational diabetic mother can deliver a healthy baby despite having diabetes. In this paper, many classification algorithms like J48, simple CART and Naïve bayes algorithm are used to diagnose the diabetes in pregnant women and they are compared for their accuracy levels.