Abstract
OBJECTIVE: To develop a nomogram prediction model for diabetic hearing impairment (DHI) based on peripheral blood long noncoding RNA MALAT1(lncRNA MALAT1), microRNA-199b (miR-199b) and advanced glycation end products (AGEs), and to evaluate its clinical utility. These biomarkers were selected due to their established roles in diabetes-related complications, including oxidative stress, inflammation, and vascular dysfunction, which are implicated in hearing impairment. METHODS: This cross-sectional study included 318 diabetic patients without pre-existing hearing loss, who were randomly divided into a training set (n=223) and a validation set (n=95) at a ratio of 7:3. Peripheral blood levels of LncRNA MALAT1 and miR-199b were quantified using real-time fluorescence quantitative PCR (RT-qPCR), and AGEs were measured by enzyme-linked immunosorbent assay (ELISA). Pure-tone audiometry (PTA) was performed to diagnose hearing impairment. Multivariate logistic regression identified independent influencing factors, and a prediction model in the form of nomogram is constructed accordingly. In order to evaluate the prediction accuracy of the model, the operating characteristic curve (ROC curve) and calibration curve of the subjects were further drawn. Decision curve analysis (DCA) was adopted to comprehensively evaluate the significance and value of the nomogram model. RESULTS: There was no significant difference in baseline characteristics between training and validation sets (all P>0.05). In the training set, diabetic peripheral neuropathy, high levels of AGEs, fasting blood glucose (FBG), postprandial blood glucose (2hPG), high expression of lncRNA MALAT1, low expression of miR-199b and long course of diabetes were independent risk factors for diabetic hearing impairment (all P<0.05). The nomogram exhibited good discriminative ability, with areas under the curve (AUC) of ROC curve of 0.810 (95% CI: 0.737-0.883) (training set) and 0.739 (95% CI: 0.597-0.882) (validation set), confirming predictive accuracy of the model. Calibration was confirmed by Hosmer-Lemeshow tests (P=0.610 and P=0.534, respectively), indicating no significant deviation from perfect fit. CONCLUSION: The nomogram integrates clinical and biomarker data to predict DHI risk, offering a practical tool for early intervention.