Performance evaluation of prediction tools for detecting individuals at high risk of type 2 diabetes in a Vietnamese population

Abstract

It is important to detect individuals at high risk of type 2 diabetes in prevention and management of the disease. The 5-year prospective study aimed at evaluating the performance of prediction tools (risk score, nomogram, prediction model) in a Vietnamese population aged 40 - 64 years (n = 1448) without diabetes at baseline. The findings showed that the performance of prediction tools developing in other populations resulted in a moderate discrimination (AUC < 0.65), sensitivity and specificity (< 0.66). It is crucial to develop prediction tools suitable for Vietnamese population.