Laboratory parameters-based logistic regression models for rapid screening of thyroid nodules

Area of Science:

• Endocrinology
• Medical Diagnostics
• Biochemistry

Background:

• Increasing incidence of thyroid nodules (TNs) presents a diagnostic challenge.
• Radiologists face mounting pressure due to the prevalence of TNs.
• Need for effective methods to differentiate benign and malignant TNs.

Purpose of the Study:

• To evaluate the effectiveness of laboratory parameters in detecting benign and malignant TNs.
• To develop early diagnosis logistic regression models using laboratory parameters.
• To assess the predictive value of these models for TN diagnosis.

Main Methods:

• Logistic regression analysis (univariate and multivariate) was performed.
• Laboratory parameters including glucose, lipid profiles, and thyroid function tests were analyzed.
• Models were developed and validated using sensitivity, specificity, and ROC curve analysis.

Main Results:

• Diagnostic models were established using key laboratory factors like HDL-C, glucose, TT3, FT3, sTSH, Lp(a), and Anti-Tg.
• Models demonstrated high AUC values for predicting benign and malignant TNs in healthy individuals.
• A model for distinguishing between benign and malignant TNs showed a relatively low AUC.

Conclusions:

• Thyroid nodules are associated with laboratory indexes related to glucose and lipid metabolism, thyroid function, albumin, MCH, and PLT.
• Laboratory parameters-based logistic regression models are recommended for routine physical examination and early screening of TNs.
• These models offer a valuable tool for early detection and management of thyroid nodules.