Related Experiment Video
Updated: Sep 19, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
2.0K
Combining structural equation modeling analysis with machine learning for early malignancy detection in Bethesda
Zeliha Aydın Kasap1, Burçin Kurt1, Ali Güner2
1Karadeniz Technical University, Faculty of Medicine, Department of Biostatistics and Medical Informatics, Trabzon, Turkey.
Artificial Intelligence in Medicine
|June 5, 2025
Summary
This study developed a decision support system using machine learning and structural equation modeling to predict malignancy in thyroid nodules with Atypia of Undetermined Significance (AUS). The system accurately identifies benign cases, reducing unnecessary surgeries.
Area of Science:
- Endocrinology
- Oncology
- Medical Informatics
Background:
- Atypia of Undetermined Significance (AUS) poses diagnostic challenges in thyroid nodule evaluation.
- Accurate preoperative risk stratification is crucial to prevent unnecessary surgical interventions.
Purpose of the Study:
- To develop a clinical decision support system for predicting malignancy in AUS thyroid nodules.
- Integrate diverse data including clinical, ultrasonography, cytopathological, and morphometric variables.
Main Methods:
- Retrospective cohort study (2011-2019) with 204 thyroid nodules from 183 patients.
- Utilized Structural Equation Modeling (SEM) for feature selection and risk factor identification.
- Employed machine learning algorithms (SVM, Naive Bayes, Decision Trees) for malignancy prediction.
Main Results:
- The Support Vector Machine (SVM) model, after SEM feature selection, achieved 82% accuracy, 97% specificity, and 84% AUC.
- SEM effectively identified risk factors for early thyroid cancer detection.
- Comparative analysis of different machine learning models was performed.
Conclusions:
- The developed model enhances clinical decision-making for AUS thyroid nodules.
- Effective identification of benign cases reduces surgical risks and improves patient care.
- The system offers a valuable tool for preoperative malignancy prediction.
