IDENTIFYING NEWLY DIAGNOSED DIABETES MELLITUS RISK FACTORS USING GRAPHICAL NETWORKS WITH EXPERT KNOWLEDGE

An increasing trend of newly diagnosed diabetes mellitus (DM) among adolescents is occurring worldwide, including Malaysia. This study aims to determine the overall relationships between risk factors on the prevalence of newly diagnosed diabetes mellitus among Malaysian adolescents. Current study uses a cross-sectional study, data from the Fifth National Health and Morbidity Survey 2015 which  consists of individuals who ages 18 and above, extraction of 18 and 19  years old from the data set was done.. Bayesian networks modelling was performed by using graphical networks with expert knowledge to identify the risk factors of newly diagnosed diabetes mellitus among adolescents in Malaysia. Education levels, Body Mass Index (BMI), and physical inactivity were identified as the significant predictors of newly diagnosed DM. The highest conditional probability of developing newly diagnosed DM belongs to both obese and underweight respondents given they have no formal education (probability, pr = 0.5000), followed by obese respondents who had unclassified level of education (pr = 0.4709), and obese respondents with primary level of education (pr = 0.4692). The findings of current study provide insights and allows policymakers to plan for future interventions in order to monitor and reduce the high prevalence of newly diagnosed diabetes mellitus among adolescents in Malaysia.