Early Stage Prediction of Type 2 Diabetes Mellitus using Prominent Genemarkers

With over 537 million people suffering from Type 2 diabetes (T2D), the global rise of this disease is quickly becoming a public health concern. International Diabetes Federation (IDF) projects the cases rising potentially up to 643 million in by 2030 and 783 million by 2045. T2D accounts for more than 90% of all diabetes cases, illustrating the need for early detection due the possible adverse effects that remain unaddressed. Basing on cell and tissue samples obtained from patients, the proposed model presents a classification model for predicting T2D predisposition using the gene expression values that have been normalized using VSN technique. Knapsack algorithm is then run on the MRMR and BDSE filtered gene markers to narrow it down to 10 most significant biomarkers. Statistical analysis confirms these biomarkers with a p-value of 0.05 are accurate. SVM was implemented with 95% accuracy for same set of genemarkers. The researchers used Microarray datasets GPL6102 and GPL570, containing 64 and 39 patients respectively, for training and testing the model. These datasets comprised 368,754 gene markers associated with diabetes. This approach is accurate and efficient for early identification of risk toward T2D, which would help millions of people change their lifestyle to avoid T2D and further help in identifying how prominent genemarkers afflict and change for different microarray technologies.