Machine learning on thyroid disease: a review

Abstract

This study reviews the recent progress of machine learning for the early diagnosis of thyroid disease. Based on the results of this review, different machine learning methods would be appropriate for different types of data for the early diagnosis of thyroid disease: (1) the random forest and gradient boosting in the case of numeric data; (2) the random forest in the case of genomic data; (3) the random forest and the ensemble in the case of radiomic data; and (4) the random forest in the case of ultrasound data. Their performance measures varied within 64.3–99.5 for accuracy, 66.8–90.1 for sensitivity, 61.8–85.5 for specificity, and 64.0–96.9 for the area under the receiver operating characteristic curve. According to the findings of this review, indeed, the following attributes would be important variables for the early diagnosis of thyroid disease: clinical stage, marital status, histological type, age, nerve injury symptom, economic income, surgery type [the quality of life 3 months after thyroid cancer surgery]; tumor diameter, symptoms, extrathyroidal extension [the local recurrence of differentiated thyroid carcinoma]; RNA feasures including ADD3-AS1 (downregulation), MIR100HG (downregulation), FAM95C (downregulation), MORC2-AS1 (downregulation), LINC00506 (downregulation), ST7-AS1 (downregulation), LOC339059 (downregulation), MIR181A2HG (upregulation), FAM181A-AS1 (downregulation), LBX2-AS1 (upregulation), BLACAT1 (upregulation), hsa-miR-9-5p (downregulation), hsa-miR-146b-3p (upregulation), hsa-miR-199b-5p (downregulation), hsa-miR-4709-3p (upregulation), hsa-miR-34a-5p (upregulation), hsa-miR-214-3p (downregulation) [papillary thyroid carcinoma]; gut microbiota RNA features such as veillonella, paraprevotella, neisseria, rheinheimera [hypothyroidism]; and ultrasound features, i.e., wreath-shaped feature, micro-calcification, strain ratio [the malignancy of thyroid nodules].

Deciphering the endometrial immune landscape of RIF during the window of implantation from cellular senescence by integrated bioinformatics analysis and machine learning

“…The RF algorithm is an ensemble method that combines many decision trees and makes a single decision on behalf of the ensemble by combing the results of multiple classifiers together ( 35 ). Each decision tree in the forest is built by using the bootstrap technique to select various samples from the original dataset and then training it with a feature set chosen by the random bagging mechanism ( 36 ). Decisions made by a large number of distinct individual trees are then voted on, and the class with the most votes as a result of the voting is assigned as the class prediction ( 37 ).…”

The progress of radiomics in thyroid nodules

“…Other information can combine the above features with building models and databases to develop classifiers to predict results, which is the ultimate goal of radiomics ( 19 ). At present, the commonly used modeling machine learning algorithms in radiomics are as follows ( 20 ): Logistic regression, random forest (RM), support vector machine (SVM), Decision Tree, k-Nearest Neighbor(KNN), artificial neural networks (ANNs), Bayesian algorithm (Bayes), clustering algorithm (such as K-means/DBSCN). All of them are supervised learning algorithms except the clustering algorithm, which is unsupervised learning ( 21 ).…”

Multi-Omics and Management of Follicular Carcinoma of the Thyroid

“…The collected information may be categorized into biographic, life-style, environmental, clinical, radiological, biochemical, cytopathological, histopathological, or genetic categories [ 169 ]. The big data that is obtained during whole genome sequencing adds to the enormity of the collected information [ 170 , 171 , 172 , 173 , 174 , 175 ]. Currently, a minute fraction of the information that is generated during the investigation is used to guide decision-making during the care of patients who have cancer, including FTC.…”

“…Its heterogeneity is manifested in its clinical presentation, findings on diagnostic investigation, response to treatment, and long-term outcome [ 35 , 150 ]. A vast amount of information is collected during history taking, physical examination, diagnostic and staging investigations, and monitoring following definitive management [ 170 , 173 ]. Analysis of the findings is influenced by the level of expertise, which explains the often very high intra- and intra-observer variation regarding histological confirmation of the type and subtypes of TC, differentiation, and changes in the TME [ 227 , 228 ].…”