Researchers have developed a nomogram designed to evaluate the probability of thyroid nodule detection among patients diagnosed with type 2 diabetes mellitus. Based on retrospective clinical data, the predictive tool incorporates six independent clinical and biochemical indicators identified through statistical modeling to estimate individual probability scores.
The investigation retrospectively evaluated a total of 320 patients with type 2 diabetes mellitus who received care at the hospital between February 2023 and January 2025. This cohort was randomly partitioned into a primary training set comprising 70 percent of the participants, corresponding to 224 individuals, and an internal validation set representing the remaining 30 percent, consisting of 96 patients. Comparative assessments confirmed that baseline clinical characteristics showed no statistically significant differences between the training cohort and the internal validation cohort, with P values exceeding 0.05.
Within the training cohort of 224 participants, baseline thyroid ultrasonography examinations were utilized to categorize subjects into distinct subgroups. Ultrasonographic screening identified 159 patients without thyroid nodules and 65 patients presenting with detected thyroid nodules. The investigators subsequently applied least absolute shrinkage and selection operator regression alongside multivariable logistic regression analyses to screen candidate variables and construct the nomogram model.
Multivariable logistic regression analysis within the training cohort established six independent factors associated with thyroid nodule detection, each maintaining statistical significance with P values below 0.05. Older age demonstrated an odds ratio of 1.122, with a 95 percent confidence interval ranging from 1.068 to 1.178, while female sex exhibited an odds ratio of 4.059, with a confidence interval of 1.603 to 10.276. Increased body mass index was independently linked to nodule detection, yielding an odds ratio of 1.520 with a 95 percent confidence interval of 1.238 to 1.866.
The statistical model further identified underlying hypertension and glycemic and endocrine biomarkers as significant independent correlates. Diagnosed hypertension was associated with an odds ratio of 5.283, with a 95 percent confidence interval between 2.045 and 13.646. Elevated glycated hemoglobin levels corresponded to an odds ratio of 2.197, with a confidence interval spanning 1.292 to 3.734, while higher insulin-like growth factor 1 concentrations showed an odds ratio of 1.019, with a 95 percent confidence interval from 1.006 to 1.032.
Evaluation of the nomogram demonstrated strong discriminatory capacity and calibration across both patient cohorts. The area under the receiver operating characteristic curve reached 0.934, with a 95 percent confidence interval of 0.899 to 0.969 in the training cohort, and 0.925, with a 95 percent confidence interval of 0.886 to 0.964 in the validation cohort. Corresponding sensitivity was measured at 84.62 percent in the training group and 80.77 percent in the validation group, alongside specificities of 86.16 percent and 87.14 percent, respectively. Good model calibration was substantiated by Hosmer-Lemeshow test P values of 0.382 for the training cohort and 0.349 for the validation cohort, evaluated alongside calibration curves and decision curve analysis. The authors concluded that the nomogram may assist in identifying patients with type 2 diabetes who face a heightened probability of thyroid nodules, while emphasizing that external validation remains necessary prior to clinical implementation.
Subject of Research: Prediction model for thyroid nodules in patients with type 2 diabetes mellitus
Article Title: Analysis of factors associated with thyroid nodules in type 2 diabetes mellitus patients and construction of a prediction model
Article References: Yao, W., & Lu, Y. (2026). Analysis of factors associated with thyroid nodules in type 2 diabetes mellitus patients and construction of a prediction model. Frontiers in Medicine, 13, Article 1930968. https://doi.org/10.3389/fmed.2026.1930968
Image Credits: AI Generated
DOI: 10.3389/fmed.2026.1930968
Keywords: type 2 diabetes mellitus, thyroid nodules, nomogram, prediction model, thyroid ultrasonography, logistic regression, Lasso regression, ROC curves, Hosmer-Lemeshow test, decision curve analysis, HbA1c, IGF-1, hypertension, body mass index, internal validation, calibration curves
News Source: Ophelia Keating. (November 18, 2026). Nomogram Predicts Thyroid Nodule Probability in Type 2 Diabetes Patients. Scienmag.



