Machine learning for risk stratification of thyroid cancer patients: a 15-year cohort study

Optimizing unsupervised feature engineering and classification pipelines for differentiated thyroid cancer recurrence prediction

BACKGROUND

Differentiated thyroid cancer (DTC) is a common endocrine malignancy with rising incidence and frequent recurrence, despite a generally favorable prognosis. Accurate recurrence prediction is critical for guiding post-treatment strategies. This study aimed to enhance predictive performance by refining feature engineering and evaluating a diverse ensemble of machine learning models using the UCI DTC dataset.

METHODS

Unsupervised data engineering-specifically dimensionality reduction and clustering-was used to improve feature quality. Principal Component Analysis (PCA) and Truncated Singular Value Decomposition (t-SVD) were selected based on superior clustering metrics: adjusted Rand Index (ARI > 0.55) and V-measure (> 0.45). These were integrated into classification pipelines using Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbors (KNN), Feedforward Neural Network (FNN), and Gradient Boosting (GB). Model performance was evaluated through bootstrapping on an independent test set, stratified 10-fold cross-validation (CV), and subgroup analyses. Metrics included balanced accuracy, F1 score, AUC, sensitivity, specificity, and precision, each reported with 95% confidence intervals (CIs). SHAP analysis supported model interpretability.

RESULTS

The PCA-based LR pipeline achieved the best test set performance: balanced accuracy of 0.95 (95% CI: 0.90-0.99), AUC of 0.99 (95% CI: 0.97-1.00), and sensitivity of 0.94 (95% CI: 0.84-1.00). In stratified CV, it maintained strong results (balanced accuracy: 0.86; AUC: 0.97; sensitivity: 0.80), with consistent performance across clinically relevant subgroups. The t-SVD-based LR pipeline showed comparable performance on both test and CV sets. SVM and FNN pipelines also performed robustly (test AUCs > 0.99; CV AUCs > 0.96). RF and KNN had high specificity but slightly lower sensitivity (test: ~0.87; CV: 0.77-0.80). GB pipelines showed the lowest overall performance (test balanced accuracy: 0.86-0.88; CV: 0.85-0.88).

CONCLUSIONS

Dimensionality reduction via PCA and t-SVD significantly improved model performance, particularly for LR, SVM, FNN, RF and KNN classifiers. The PCA-based LR pipeline showed the best generalizability, supporting its potential integration into clinical decision-support tools for personalized DTC management.

CLINICAL TRIAL NUMBER

Not applicable.

Role of artificial intelligence in early identification and risk evaluation of non-communicable diseases: a bibliometric analysis of global research trends

OBJECTIVE

This study aims to shed light on the transformative potential of artificial intelligence (AI) in the early detection and risk assessment of non-communicable diseases (NCDs).

STUDY DESIGN

Bibliometric analysis.

SETTING

Articles related to AI in early identification and risk evaluation of NCDs from 2000 to 2024 were retrieved from the Scopus database.

METHODS

This comprehensive bibliometric study focuses on a single database, Scopus and employs narrative synthesis for concise yet informative summaries. Microsoft Excel V.365 and VOSviewer software (V.1.6.20) were used to summarise bibliometric features.

RESULTS

The study retrieved 1745 relevant articles, with a notable surge in research activity in recent years. Core journals included Scientific Reports and IEEE Access, and core institutions included the Harvard Medical School and the Ministry of Education of the People's Republic of China, while core countries comprised China, the USA, India, the UK and Saudi Arabia. Citation trends indicated substantial growth and recognition of AI's impact on NCDs management. Frequent author keywords identified key research hotspots, including specific NCDs like Alzheimer's disease and diabetes. Risk assessment studies demonstrated improved predictions for heart failure, cardiovascular risk, breast cancer, diabetes and inflammatory bowel disease.

CONCLUSION

Our findings highlight the increasing role of AI in early detection and risk prediction of NCDs, emphasising its widening research impact and future clinical potential.

Erosion of the temporal bone by vestibular schwannoma: morphometrics and predictive modeling

PURPOSE

To perform a comprehensive morphometric analysis of vestibular schwannomas (VS) using multimodal imaging, focusing on the relationship between tumor characteristics and internal acoustic canal (IAC) changes.

METHODS

We analyzed a cohort of patients undergoing radiosurgery for VS, utilizing high-definition MRI and bone CT for detailed anatomical assessment. Image co-registration and fusion techniques were employed to examine VS and IAC dimensions. Advanced statistical methods, including logistic regression, were applied to identify patterns of IAC dilation and establish predictive indicators for these morphological changes.

RESULTS

The study included 659 patients (51.1% female, mean age 56.37 years) with evenly distributed tumor lateralization. Koos grades were I (22.9%), II (29.9%), III (38.2%), IVa (8.9%), and IVb (0.3%). Most tumors (90.9%) extended both inside and outside the IAC. Ipsilateral IAC (IIAC) dimensions were significantly larger than contralateral, with IIAC volume 49% greater (p < .0001). Higher Koos grades correlated with increased intra-canalicular lesion volume (icLV), which was strongly associated with IIAC size. Logistic regression identified icLV as the strongest predictor of IIAC dilation (AUC = 0.7674, threshold = 137.52 mm3).

CONCLUSION

The icLV appears central to the pathophysiological development of VS and its impact on IAC anatomy. While limited by a selective patient base and static imaging data, these findings enhance the understanding of VS-IAC interactions, offering insights for improved clinical management and further research.

Thy-DAMP: deep artificial neural network model for prediction of thyroid cancer mortality

PURPOSE

Despite the rising incidence of differentiated thyroid cancer (DTC), mortality rates have remained relatively low yet crucial for effective patient management. This study aims to develop a deep neural network capable of predicting mortality in patients with differentiated thyroid cancer.

METHODS

Leveraging data from the Surveillance, Epidemiology, and End Results (SEER) database, we developed Thy-DAMP (Deep Artificial Neural Network Model for Prediction of Thyroid Cancer) to forecast mortality in DTC patients. The dataset comprised demographic, histologic, and staging information. Following data normalization and feature encoding, the dataset was partitioned into training, testing, and validation subsets, with model hyperparameters fine-tuned via cross-validation.

RESULTS

Among the 63,513 patients, the mean age was 48.22 years (SD = 14.8), with 77.32% being female. Papillary carcinoma emerged as the predominant subtype, representing 62.94% of cases. The majority presented with stage I disease (73.96%). Thy-DAMP demonstrated acceptable performance metrics on both the test and validation sets. Sensitivity was 83.24% (95% CI 76.95-88.40%), specificity was 93.53% (95% CI 93.01-94.02%), and accuracy stood at 93.33% (95% CI 92.82-93.83%). The model exhibited a positive predictive value of 19.76% (95% CI 18.20-21.42%) and a negative predictive value of 99.66% (95% CI 99.53-99.75%). Additionally, Thy-DAMP demonstrated a positive likelihood ratio of 12.86 (95% CI 11.62-14.23), a negative likelihood ratio of 0.18 (95% CI 0.13-0.25), and an area under the receiver operating characteristic curve (AUROC) of 0.95. The model was externally validated on a separate dataset with nearly identical performance.

CONCLUSION

Thy-DAMP showcases considerable promise in accurately predicting mortality in DTC patients, leveraging limited set of patient data.

Explainable Thyroid Cancer Diagnosis Through Two-Level Machine Learning Optimization with an Improved Naked Mole-Rat Algorithm

Modern technologies, particularly artificial intelligence methods such as machine learning, hold immense potential for supporting doctors with cancer diagnostics. This study explores the enhancement of popular machine learning methods using a bio-inspired algorithm-the naked mole-rat algorithm (NMRA)-to assess the malignancy of thyroid tumors. The study utilized a novel dataset released in 2022, containing data collected at Shengjing Hospital of China Medical University. The dataset comprises 1232 records described by 19 features. In this research, 10 well-known classifiers, including XGBoost, LightGBM, and random forest, were employed to evaluate the malignancy of thyroid tumors. A key innovation of this study is the application of the naked mole-rat algorithm for parameter optimization and feature selection within the individual classifiers. Among the models tested, the LightGBM classifier demonstrated the highest performance, achieving a classification accuracy of 81.82% and an F1-score of 86.62%, following two-level parameter optimization and feature selection using the naked mole-rat algorithm. Additionally, explainability analysis of the LightGBM model was conducted using SHAP values, providing insights into the decision-making process of the model.

Accurate diagnosis of acute appendicitis in the emergency department: an artificial intelligence-based approach

The diagnosis of abdominal pain in emergency departments is challenging, and appendicitis is a common concern. Atypical symptoms often delay diagnosis. Although the Alvarado score aids in decision-making, its low specificity can lead to unnecessary surgeries. By leveraging machine learning, we aim to enhance diagnostic accuracy by predicting appendicitis and distinguishing it from other causes of abdominal pain in the emergency department. Data were collected from 534 patients who presented with acute abdominal pain. Patient characteristics, laboratory results, and causes of pain were recorded. Machine learning algorithms (support vector classifier, random forest classifier, gradient boosting classifier, and Gaussian naive Bayes) were used to predict the cause of pain. Model calibration was assessed using the Brier score. The mean age was 46.89 (20.3) years, with an almost equal sex distribution (49% male, 51% female). Cholecystitis was the most prevalent outcome (37.07%), followed by appendicitis (25.84%). The Gaussian naive Bayes model exhibited superior performance in terms of accuracy (95.03% 95% CI 90.44-97.83%), sensitivity (87.18% 95% CI 72.57-95.70%), and specificity (97.54% 95% CI 92.98-99.49%), while the random forest model showed a sensitivity of 79.49%, specificity of 96.72%, and accuracy of 92.55%. The gradient boosting algorithm achieved a sensitivity, specificity, and accuracy of 89.74%, 95.90%, and 94.41%, respectively. The support vector classifier demonstrated a sensitivity of 89.74%, specificity of 92.62%, and accuracy of 91.93%. The use of modern machine learning methods aids in the accurate diagnosis of appendicitis.

Deep radiomics-based prognostic prediction of oral cancer using optical coherence tomography

BACKGROUND

This study aims to evaluate the integration of optical coherence tomography (OCT) and peripheral blood immune indicators for predicting oral cancer prognosis by artificial intelligence.

METHODS

In this study, we examined patients undergoing radical oral cancer resection and explored inherent relationships among clinical data, OCT images, and peripheral immune indicators for oral cancer prognosis. We firstly built a peripheral blood immune indicator-guided deep learning feature representation method for OCT images, and further integrated a multi-view prognostic radiomics model incorporating feature selection and logistic modeling. Thus, we can assess the prognostic impact of each indicator on oral cancer by quantifying OCT features.

RESULTS

We collected 289 oral mucosal samples from 68 patients, yielding 1,445 OCT images. Using our deep radiomics-based prognosis model, it achieved excellent discrimination for oral cancer prognosis with the area under the receiver operating characteristic curve (AUC) of 0.886, identifying systemic immune-inflammation index (SII) as the most informative feature for prognosis prediction. Additionally, the deep learning model also performed excellent results with 85.26% accuracy and 0.86 AUC in classifying the SII risk.

CONCLUSIONS

Our study effectively merged OCT imaging with peripheral blood immune indicators to create a deep learning-based model for inflammatory risk prediction in oral cancer. Additionally, we constructed a comprehensive multi-view radiomics model that utilizes deep learning features for accurate prognosis prediction. The study highlighted the significance of the SII as a crucial indicator for evaluating patient outcomes, corroborating our clinical statistical analyses. This integration underscores the potential of combining imaging and blood indicators in clinical decision-making.

TRIAL REGISTRATION

The clinical trial associated with this study was prospectively registered in the Chinese Clinical Trial Registry with the trial registration number (TRN) ChiCTR2200064861. The registration was completed on 2021.

Analysis of risk factors for papillary thyroid carcinoma and the association with thyroid function indicators

Objective

This study aims to analyze the relationship between papillary thyroid carcinoma (PTC) and various factors.

Methods

The study involved two groups-PTC patients and non-PTC controls. We utilized binary logistic regression and Least Absolute Shrinkage and Selection Operator (Lasso) regression for variable selection and risk factor analysis. Correlation analysis was performed using Spearman's rank correlation. The diagnostic value of thyroid stimulating hormone (TSH) levels for PTC was assessed using Receiver Operating Characteristic (ROC) curves.

Results

PTC patients exhibited higher body mass index (BMI) (23.71 vs. 22.66, p<0.05) and TSH levels (3.38 vs. 1.59, p<0.05). Urinary iodine concentration (UIC) was an independent predictor of PTC (OR=1.005, p<0.05). The optimal TSH threshold for PTC diagnosis was 2.4 mIU/L [The Area Under the Curve (AUC)=67.3%, specificity=71.4%, sensitivity=70.1%]. TSH levels positively correlated with BMI (r=0.593, p<0.05) and UIC (r=0.737, p<0.05).

Conclusions

UIC may be an independent predictor of PTC, and TSH levels have some diagnostic value for identifying PTC.

Air Traffic Controller Workload Detection Based on EEG Signals

The assessment of the cognitive workload experienced by air traffic controllers is a complex and prominent issue in the research community. This study introduces new indicators related to gamma waves to detect controllers' workload and develops experimental protocols to capture their EEG data and NASA-TXL data. Then, statistical tests, including the Shapiro-Wilk test and ANOVA, were used to verify whether there was a significant difference between the workload data of the controllers in different scenarios. Furthermore, the Support Vector Machine (SVM) classifier was employed to assess the detection accuracy of these indicators across four categorizations. According to the outcomes, hypotheses suggesting a strong correlation between gamma waves and an air traffic controller's workload were put forward and subsequently verified; meanwhile, compared with traditional indicators, the indicators associated with gamma waves proposed in this paper have higher accuracy. In addition, to explore the applicability of the indicator, sensitive channels were selected based on the mRMR algorithm for the indicator with the highest accuracy, β + θ + α + γ, showcasing a recognition rate of a single channel exceeding 95% of the full channel, which meets the requirements of convenience and accuracy in practical applications. In conclusion, this study demonstrates that utilizing EEG gamma wave-associated indicators can offer valuable insights into analyzing workload levels among air traffic controllers.

Development and validation of prediction models for papillary thyroid cancer structural recurrence using machine learning approaches

BACKGROUND

Although papillary thyroid cancer (PTC) patients are known to have an excellent prognosis, up to 30% of patients experience disease recurrence after initial treatment. Accurately predicting disease prognosis remains a challenge given that the predictive value of several predictors remains controversial. Thus, we investigated whether machine learning (ML) approaches based on comprehensive predictors can predict the risk of structural recurrence for PTC patients.

METHODS

A total of 2244 patients treated with thyroid surgery and radioiodine were included. Twenty-nine perioperative variables consisting of four dimensions (demographic characteristics and comorbidities, tumor-related variables, lymph node (LN)-related variables, and metabolic and inflammatory markers) were analyzed. We applied five ML algorithms-logistic regression (LR), support vector machine (SVM), extreme gradient boosting (XGBoost), random forest (RF), and neural network (NN)-to develop the models. The area under the receiver operating characteristic (AUC-ROC) curve, calibration curve, and variable importance were used to evaluate the models' performance.

RESULTS

During a median follow-up of 45.5 months, 179 patients (8.0%) experienced structural recurrence. The non-stimulated thyroglobulin, LN dissection, number of LNs dissected, lymph node metastasis ratio, N stage, comorbidity of hypertension, comorbidity of diabetes, body mass index, and low-density lipoprotein were used to develop the models. All models showed a greater AUC (AUC = 0.738 to 0.767) than did the ATA risk stratification (AUC = 0.620, DeLong test: P < 0.01). The SVM, XGBoost, and RF model showed greater sensitivity (0.568, 0.595, 0.676), specificity (0.903, 0.857, 0.784), accuracy (0.875, 0.835, 0.775), positive predictive value (PPV) (0.344, 0.272, 0.219), negative predictive value (NPV) (0.959, 0.959, 0.964), and F1 score (0.429, 0.373, 0.331) than did the ATA risk stratification (sensitivity = 0.432, specificity = 0.770, accuracy = 0.742, PPV = 0.144, NPV = 0.938, F1 score = 0.216). The RF model had generally consistent calibration compared with the other models. The Tg and the LNR were the top 2 important variables in all the models, the N stage was the top 5 important variables in all the models.

CONCLUSIONS

The RF model achieved the expected prediction performance with generally good discrimination, calibration and interpretability in this study. This study sheds light on the potential of ML approaches for improving the accuracy of risk stratification for PTC patients.

TRIAL REGISTRATION

Retrospectively registered at www.chictr.org.cn (trial registration number: ChiCTR2300075574, date of registration: 2023-09-08).