Risk Prediction Model for Elderly Differentiated Thyroid Cancer Based on Combined Sleep Quality Assessment and Multimodal Ultrasound

TL;DR

This study creates a risk prediction model for elderly thyroid cancer by combining sleep quality and ultrasound data to improve preoperative diagnosis.

Contribution

The novel contribution is integrating sleep quality assessment with multimodal ultrasound to enhance thyroid cancer risk prediction in elderly patients.

Findings

The combined model achieved an AUC of 0.860, outperforming models using ultrasound alone or with TPOAB.

Key risk factors included sleep quality (PSQI > 7), nodule shape, calcification, and blood flow.

The nomogram model showed strong discrimination and clinical utility in validation tests.

Abstract

To explore the differential diagnosis for benign and malignant thyroid nodules and the diagnostic value of sleep quality, to construct and validate a risk prediction model, providing the basis for clinical treatment decision for elderly thyroid cancer.

Clinical data, Pittsburgh Sleep Quality Index (PSQI), and multimodal ultrasound were collected from elderly patients undergoing fine needle aspiration biopsy or thyroid surgery in our department of endocrinology and general surgery. Postoperative pathological results served as the gold standard, binary logistic regression identified significant risk factors, and the receiver‐operating characteristic (ROC) curves were plotted to construct and validate the prediction model.

Among 763 enrolled patients (566 benign and 197 malignant), multivariate analysis revealed independent risk factors: TPOAB positive, daytime dysfunction, PSQI > 7, irregular nodule shape, calcification, blood flow, high elasticity scores, and low contrast enhancement. The area under the curve (AUC) for the combined model was 0.860, significantly higher than models using multimodal ultrasound alone (AUC = 0.824) or multimodal ultrasound with TPOAB (AUC = 0.831), p < 0.05. The nomogram‐based prediction model demonstrated excellent discrimination, calibration, and clinical utility in internal and external validation.

Integrating sleep quality assessment with multimodal ultrasound assisted in the differentiation of thyroid nodules in the elderly, thus may improve the preoperative diagnostic levels. Risk prediction model in a nomogram format provided an intuitive and reliable tool for clinical decision‐making.

Sleep disorders may promote the occurrence of tumours, disrupt the rhythm, and delay the expression time of cancer‐related genes, make susceptible individuals more prone to DNA damage, and reduce the efficiency of cell repair as well. Therefore, we evaluated the individualized risk of participants through clinical data, sleep quality assessment, and multimodal ultrasound. Then, we constructed an economical, convenient, and operable prediction model to guide FNA.

Figures