Abstract
Diabetes mellitus is a global significant health concern that affects various body systems. Current assessment tools mostly rely on individual biomarkers, which do not fully capture the overall health burden of the disease. Therefore, a comprehensive and understandable measure is needed to evaluate the health impact associated with diabetes. This study aimed to develop the Diabetic Health Impact Indicator (DHII), a multidimensional, biomarker-based summary score designed to reflect the overall effects of diabetes, by employing a cross-sectional design with secondary data from 1,000 individuals, including diabetic, prediabetic, and non-diabetic participants. The dataset encompassed biochemical, physiological, and demographic variables related to diabetes. Exploratory Factor Analysis (EFA) was utilized to identify underlying health domains, with factor loadings used to assign weights. The empirical cumulative distribution function (ECDF) scaled the composite scores within a 0-10 range. Gaussian mixture modelling showed that DHII followed heterogeneous distribution patterns across subgroups, with K-CGMD providing the best fit in most cases. Predictive validation using logistic regression showed that DHII alone achieved an AUC of 0.721, while a baseline model with age, sex, and BMI achieved an AUC of 0.961. Adding DHII to the baseline model resulted in a marginal increase in AUC to 0.968, which was not statistically significant (DeLong’s test, p=0.33). In summary, DHII is a data-driven composite index with multiple biomarker domains that demonstrates moderate discriminatory ability for diabetes status. Further validation in independent populations is required to assess its clinical use.