Risk factors and diagnostic prediction models for papillary thyroid carcinoma

Abstract

Thyroid nodules (TNs) represent a common scenario. More accurate pre-operative diagnosis of malignancy has become an overriding concern. This study incorporated demographic, serological, ultrasound, and biopsy data and aimed to compare a new diagnostic prediction model based on Back Propagation Neural Network (BPNN) with multivariate logistic regression model, to guide the decision of surgery. Records of 2,090 patients with TNs who underwent thyroid surgery were retrospectively reviewed. Multivariate logistic regression analysis indicated that Bethesda category (OR=1.90, P<0.001), TIRADS (OR=2.55, P<0.001), age (OR=0.97, P=0.002), nodule size (OR=0.53, P<0.001), and serum levels of Tg (OR=0.994, P=0.004) and HDL-C (OR=0.23, P=0.001) were statistically significant independent differentiators for patients with PTC and benign nodules. Both BPNN and regression models showed good accuracy in differentiating PTC from benign nodules (area under the curve [AUC], 0.948 and 0.924, respectively). Notably, the BPNN model showed a higher specificity (88.3% vs. 73.9%) and negative predictive value (83.7% vs. 45.8%) than the regression model, while the sensitivity (93.1% vs. 93.9%) was similar between two models. Stratified analysis based on Bethesda indeterminate cytology categories showed similar findings. Therefore, BPNN and regression models based on a combination of demographic, serological, ultrasound, and biopsy data, all of which were readily available in routine clinical practice, might help guide the decision of surgery for TNs.

Do metabolic factors increase the risk of thyroid cancer? a Mendelian randomization study

“…HDL-C level was found to be a statistically significant independent predictor of thyroid cancer in a model developed by Zhang et al. ( 32 ). Some retrospective observational studies have reported an association between total cholesterol ( 31 ) and apolipoprotein A1 ( 33 ) with thyroid cancer, which is somewhat inconsistent with the results of our study.…”

“…Similarly, the Swedish Apolipoprotein-Related Mortality Risk (AMORIS) Cohort study demonstrated that thyroid cancer risk was associated with blood levels of total cholesterol (TC) and HDL-C (31). HDL-C level was found to be a statistically significant independent predictor of thyroid cancer in a model developed by Zhang et al (32). Some retrospective observational studies have reported an association between total cholesterol (31) and apolipoprotein A1 (33) with thyroid cancer, which is somewhat inconsistent with the results of our study.…”

Establishment and validation of a multivariate logistic model for risk factors of thyroid nodules using lasso regression screening

“…The risk factor for thyroid nodules identified in this study is dyslipidemia, which also constitutes a component of metabolic syndrome and exhibits similarities to obesity and impaired fasting glucose. Extensive research has demonstrated that elevated TG and decreased HDL-C are associated with an increased risk of thyroid nodules ( 8 , 11 , 41 ). In this study, the levels of TG and TC in the group with thyroid nodules were significantly higher than those in the control group, while the levels of HDL-C were significantly lower.…”

“…Approximately 10-15% of thyroid nodules are malignant, signifying the presence of thyroid cancer (2)(3)(4)(5)(6)(7). The incidence of thyroid nodules and thyroid cancer has gradually increased over the past three decades (7)(8)(9)(10)(11). This increase may be attributed to advances in diagnostic tools, including high-resolution ultrasound and computerized tomography scans (12)(13)(14).…”

A risk prediction model for type 2 diabetes mellitus complicated with retinopathy based on machine learning and its application in health management

“…Therefore, there is a need for a DR risk prediction model that contains fewer indicators and is thus more accessible. Because different screening methods have different characteristics, some key risk factors may be neglected, or some features with poor prediction ability may be included in the screening process ( 25 , 31 – 34 ). For example, RF uses an integrated algorithm, which is better than most single algorithms in its accuracy, but will overfit the classification problem with high noise ( 35 , 36 ).…”