Exploring Clinical Determinants and Machine Learning–Based Prediction of Disease Outcomes in Graves Ophthalmopathy

Exploring Clinical Determinants and Machine Learning–Based Prediction of Disease Outcomes in Graves Ophthalmopathy | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Exploring Clinical Determinants and Machine Learning–Based Prediction of Disease Outcomes in Graves Ophthalmopathy Anıl Ağaçkesen, Titap Yazicioglu, Murat Oklar, Ahmet Can Özdemir, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8484257/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Graves ophthalmopathy (GO) exhibits heterogeneous clinical behavior, and predicting disease severity and course remains challenging. This study aimed to evaluate clinical determinants of GO using conventional statistical analyses and machine learning (ML) approaches to improve prognostic prediction. Methods Medical records of 153 patients with GO were retrospectively reviewed. Demographic characteristics, clinical findings, thyroid function, autoantibody levels, imaging features, and smoking status were analyzed. Patients were classified according to disease severity, activity, onset pattern, and tissue predominance. In addition to classical statistical tests, logistic regression, random forest, support vector classifier (SVC), and k-nearest neighbors (k-NN) algorithms were trained and evaluated using accuracy, F1-score, and AUC metrics. Results Male sex and active smoking were significantly associated with higher disease severity and activity. Muscle-predominant GO was associated with older age, male sex, smoking, higher severity, and higher activity. No significant differences were observed among clinical subgroups regarding thyroid hormone levels or autoantibody titers. Among ML models, the SVC showed the best performance for predicting severity (AUC = 0.818); Random Forest best estimated tissue predominance (AUC = 0.662), while logistic regression showed the highest accuracy for onset prediction. Smoking was the strongest predictor of EUGOGO-based severity, whereas age was most influential in NOSPECS-based assessments. Conclusion Demographic and clinical factors, particularly smoking, sex, age, and muscle involvement, appear to be stronger determinants of GO course than thyroid-related biochemical parameters. Machine learning approaches demonstrated meaningful discriminatory capability for predicting complex clinical outcomes and may contribute to future risk stratification frameworks. Health sciences/Risk factors Health sciences/Medical research/Outcomes research Figures Figure 1 Figure 2 INTRODUCTION Graves ophthalmopathy(GO) is the most common extrathyroidal manifestation of Graves disease, affecting the ocular adnexa and the orbit. Although it is recognized as an autoimmune inflammatory disorder, its underlying pathophysiological mechanisms have not been fully elucidated. Thyrotropin receptor antibodies are considered to play a central role in the pathogenesis; however, insulin-like growth factor-1 (IGF-1) receptors, environmental influences, and genetic factors also contribute to disease development. Clinically, the disease may present with eyelid retraction, proptosis, and diplopia, and in more severe cases may lead to sight-threatening complications such as corneal exposure and optic neuropathy. Ophthalmopathy occurs in approximately 40% of patients with Graves disease [ 1 ]. While conservative management and control of risk factors are usually sufficient in mild cases, immunosuppressive therapy, radiotherapy, or surgical intervention may be required in severe disease. Smoking, elevated thyrotropin receptor antibody levels, thyroid dysfunction, radioactive iodine therapy, and hypercholesterolemia have been reported as important risk factors [ 2 , 3 ]. In addition, selenium deficiency, female sex, and genetic predisposition have also been associated with increased risk [ 4 ]. A large-scale study conducted in a Chinese population reported that older age and male sex were associated with more severe disease [ 5 ]. Predicting whether ophthalmopathy will develop in patients with Graves disease, when it will occur, and how severe it will be remains challenging. Considerable interindividual variability in clinical presentation and disease course is thought to reflect the multifactorial and complex nature of the underlying pathophysiology. Machine learning (ML) techniques have the potential to identify complex patterns that may not be captured by conventional statistical methods and can model nonlinear interactions among multiple clinical, biochemical, and demographic variables. Integration of ML-based approaches into clinical practice may facilitate more accurate early risk stratification and prediction of disease prognosis. The aim of this study was to investigate risk factors associated with the development of ophthalmopathy in patients presenting to the ophthalmology outpatient clinic using both conventional statistical analyses and ML-based methods, and to contribute to the literature, particularly in terms of prognostic prediction. MATERIALS AND METHODS Study Design Ethical approval for the study was obtained from the institutional ethics committee (Approval number: 2025/010.99/20/1). Patients diagnosed with Graves ophthalmopathy who presented to our oculoplastic clinic were retrospectively reviewed. Thyroid hormone levels (T3, T4, TSH) at the time of initial Graves disease diagnosis, thyroid functional status (hyperthyroid, euthyroid, or hypothyroid), thyroid autoantibodies (TSHR-Ab, Anti-TPO), age, and sex were recorded. In addition, the time interval between the diagnosis of Graves disease and the onset of ophthalmopathy symptoms was noted in years. Comprehensive ophthalmological examination findings were evaluated. In unilateral cases, the affected eye, and in bilateral cases, the eye with more severe involvement was included in the analysis. Upper eyelid retraction was defined as an eyelid position ≥1 mm above the limbus. Patients with constant diplopia were considered diplopia-positive, and those with marked limitation in any gaze position were considered to have motility restriction. Hertel exophthalmometry measurements were recorded, and values ≥21 mm or an interocular difference ≥2 mm were accepted as significant proptosis. Additional ophthalmic comorbidities such as dry eye and glaucoma, as well as systemic comorbidities including hypertension, diabetes mellitus, and other autoimmune diseases, were documented. History of thyroidectomy and current smoking status were also recorded. Disease severity was classified according to the European Group ON Graves Orbitopathy (EUGOGO) 2021 criteria as mild or moderate–severe; additionally, NOSPECS scores were calculated for all patients. Disease activity was assessed using the Clinical Activity Score (CAS) in accordance with the EUGOGO 2021 guidelines; CAS ≥3 was defined as active and <3 as inactive ophthalmopathy. Patients who developed ophthalmopathy simultaneously with Graves disease or within 6 months were classified as early-onset, whereas those who developed ophthalmopathy later were classified as late-onset. Orbital magnetic resonance imaging (MRI) scans were retrospectively evaluated. Based on tissue predominance, patients were classified into muscle-predominant and fat-predominant subtypes, as described in the literature. Muscle-predominant ophthalmopathy was defined as prominent enlargement and thickening of the extraocular muscles with comparatively less orbital fat expansion, whereas fat-predominant ophthalmopathy was characterized by marked orbital fat expansion with minimal or mild extraocular muscle involvement. Evaluations were performed semi-quantitatively by experienced ophthalmologists, taking radiological findings into consideration. Machine Learning Analysis All analyses were performed in Python 3.12 using the pandas and scikit-learn libraries. For continuous variables, missing values were imputed using the mean and subsequently standardized; for categorical variables, missing data were completed with the most frequent category followed by one-hot encoding. A column transformer was used to ensure that the same preprocessing procedures were consistently applied across all models. Each target variable (outcome) was analyzed independently using the full set of available predictor variables. To provide a model-independent evaluation framework, four different classifier families were compared: Logistic Regression Random Forest Classifier Support Vector Classifier (SVC with radial basis function kernel) k-Nearest Neighbors (k-NN) Hyperparameters for each model were optimized using a 30-iteration, stratified 5-fold randomized cross-validation search. To ensure full reproducibility, the random seed was fixed at 42. For binary outcomes, model optimization was guided by the area under the receiver operating characteristic curve (AUC), whereas the macro-averaged F1-score was used for multiclass endpoints. The dataset was randomly divided into training (80%) and test (20%) sets while preserving class balance. Following hyperparameter optimization, the best-performing model for each outcome was retrained on the full training dataset and subsequently evaluated on the independent test set. Model performance was summarized using accuracy, balanced accuracy, macro-F1 score, and AUC (for binary targets). ROC curves were plotted to visualize discriminative performance. To quantitatively assess the contribution of each predictor, permutation importance was calculated on the test set using 10 random permutations per feature. This model-agnostic approach quantifies the mean reduction in performance (AUC or F1-score) when the association between a given feature and the outcome is disrupted; therefore, higher importance values indicate stronger model dependency on that variable. To maintain comparability across different algorithmic families, permutation importance was preferred over internal model coefficient–based measures, even for linear and tree-based models. Feature importance results were visualized using horizontal bar plots displaying mean importance values and corresponding standard deviations. All analyses were conducted using scikit-learn version 1.5, and visualizations were generated with matplotlib . Randomness was fully synchronized across Python, NumPy, and scikit-learn to ensure deterministic results. All scripts were executed in a controlled virtual environment (Python venv) on a Windows 10 operating system. Statistical analyses were performed using SPSS software (IBM SPSS Statistics, version 23.0). Normality of continuous variables was assessed separately for each group using the Shapiro–Wilk test. Continuous variables with a normal distribution were presented as mean ± standard deviation (Mean ± SD), whereas non-normally distributed variables were expressed as median (Q1–Q3). For comparisons between two groups, the Independent Samples t-test was used when normality assumptions were met, and the Mann–Whitney U test was applied otherwise. Homogeneity of variances was evaluated using Levene’s test. Categorical variables were expressed as frequencies and percentages, and group comparisons were performed using the Pearson chi-square test or Fisher’s Exact test, when appropriate. A p-value < 0.05 was considered statistically significant. RESULTS Medical records of 153 patients (108 females, 45 males) were reviewed. The mean age of the patients was 47.45 ± 12.34 years. Patients were classified into two groups in four separate analyses according to disease activity (active / inactive), severity (mild / moderate–severe), onset time (early / late onset), and tissue involvement pattern (fat-predominant / muscle-predominant). When classified according to disease severity, the moderate–severe group demonstrated significantly higher clinical activity scores, a higher proportion of male sex, and a significantly greater rate of active smoking. When grouped according to disease activity, the active group had a significantly higher proportion of male sex and greater disease severity; however, no significant difference was observed in terms of smoking status. Additionally, the frequency of dry eye was significantly higher in the inactive group compared with the active group. With respect to clinical examination findings, eyelid retraction, diplopia, restricted ocular motility, proptosis, higher Hertel measurements, and extraocular muscle involvement were significantly more frequent both in the moderate–severe severity group and in the active disease group. Statistical comparisons according to disease severity and activity are summarized in Table 1. Table1. Statistical analysis of cases according to severity and activity. When patients were evaluated according to the onset time of ophthalmopathy, those with late-onset disease had a significantly longer follow-up duration. However, no statistically significant difference was observed between early- and late-onset groups with respect to the evaluated clinical and demographic parameters (Table 2). Regarding tissue involvement patterns, patients with muscle-predominant ophthalmopathy tended to be significantly older, more frequently male, and more likely to be active smokers. In addition, active disease and moderate–severe severity were more commonly observed in the muscle-predominant group. Clinical examination findings such as eyelid retraction, diplopia, and restricted ocular motility were also significantly more frequent in patients with muscle-predominant involvement (Table 2). Across all analyses, no statistically significant differences were detected between the groups in terms of thyroid hormone levels, thyroid functional status, or thyroid autoantibodies (Tables 1 and 2). Table2. Statistical analysis of cases according to ophthalmopathy onset and tissue dominancy. Following the application of machine learning models, the best predictive performance for ophthalmopathy severity was achieved with the Support Vector Classifier (SVC), reaching an AUC of 0.818. For ophthalmopathy onset, the highest performance was obtained with logistic regression, whereas Random Forest yielded the best performance for tissue predominance with an AUC of 0.662. NOSPECS score and disease activity were best predicted by Random Forest and k-Nearest Neighbors (k-NN), both achieving an AUC of approximately 0.64. The predictive performance levels of the clinical outcomes across different models and the corresponding ROC curves are summarized in Figure1. Feature importance analyses demonstrated that machine learning models highlighted different predictors for different clinical outcomes (Figure2). In the model classifying disease severity according to EUGOGO, smoking status emerged as the strongest predictor, whereas in the NOSPECS-based classification, age was the most influential variable, with other parameters contributing at relatively similar levels. In the model predicting disease activity, sex, thyroid hormone levels, and thyroid autoantibodies provided substantial predictive contribution. For distinguishing early- versus late-onset ophthalmopathy, age was identified as the most important determinant. In the tissue predominance model, sex and TSH receptor autoantibodies played a prominent predictive role. DISCUSSION The etiopathogenesis of Graves ophthalmopathy has not yet been fully elucidated and the disease is considered multifactorial. In addition to genetic and environmental influences, factors such as smoking, sex, age, and thyroid autoantibodies are also thought to play a role in its development and pathogenesis [6]. In the present study, the clinical characteristics of ophthalmopathy in patients with Graves disease were comprehensively evaluated using both conventional statistical methods and ML-based models. In our study, evaluation of disease severity demonstrated that male sex and active smoking were significantly associated with more severe ophthalmopathy. Similarly, the proportion of male patients was significantly higher in the active disease group. These findings are consistent with results from previous large-scale studies [7,8]. It has been suggested that a more pronounced inflammatory response and a tendency for delayed presentation to medical care in male patients may contribute to this observation. Smoking is considered one of the most important modifiable risk factors in the pathogenesis of GO [2]. It is well established that GO develops more frequently in patients with Graves disease who smoke, and that the disease follows a more severe course in smokers with GO [9]. Smoking has been reported to enhance oxidative stress and induce fibrosis-related gene expression in orbital fibroblasts [10]. Another molecular study demonstrated that smoking promotes hypoxia, thereby triggering adipogenesis and angiogenesis within the orbit [11]. In our study, the strong association between smoking and disease severity is consistent with these proposed pathophysiological mechanisms. When clinical findings related to disease activity and severity were evaluated, eyelid retraction, diplopia, restricted ocular motility, proptosis, and increased Hertel measurements were significantly more frequent in both the active and moderate–severe groups. Furthermore, an increased number of extraocular muscle involvements was strongly associated with both disease activity and severity, suggesting that orbital muscle involvement and mechanical restriction reflect more advanced stages of the disease and parallel functional deterioration. In the analysis based on tissue predominance, the muscle-predominant subtype was found to be associated with older age, male sex, and smoking, and it was accompanied by more active and more severe ophthalmopathy. Consistent with our findings, Wiersinga et al. reported greater extraocular muscle enlargement in older patients, active smokers, and those with more severe ophthalmopathy in a radiological study tissue [12]. Differences in fibroblast phenotypes, cytokine profiles, and orbital T-cell subsets have been proposed as potential cellular and molecular mechanisms contributing to the differential involvement of fat and muscle tissue [12]. In addition, studies comparing extraocular muscle volume with total orbital volume have demonstrated that increased muscle volume is associated with active orbitopathy [13]. In our study group, the finding of higher disease severity in muscle-dominant patients suggests that this subtype represents a more clinically aggressive phenotype. Collectively, these findings indicate that muscle-predominant involvement may serve not only as a morphological subclassification but also as an important prognostic indicator. Another noteworthy finding of our study is that thyroid hormone levels, thyroid functional status, and autoantibody titers did not differ significantly among any of the clinical subgroups in terms of disease severity, activity, onset pattern, or tissue predominance. This observation supports the concept that Graves ophthalmopathy is not merely a direct reflection of systemic thyroid autoimmunity; rather, local orbital immune responses, environmental influences, and individual susceptibility may play a more dominant role. From a clinical perspective, this suggests that relying solely on biochemical parameters for prognostic prediction in Graves ophthalmopathy may be insufficient. In recent years, the use of artificial intelligence and machine learning (ML)-based models in the field of thyroid ophthalmopathy has markedly increased, particularly for disease prediction and risk stratification [14]. One of the earliest examples of AI application in Graves ophthalmopathy was reported by Salvi et al. in 2002, where an artificial neural network was employed to predict disease progression at the time of initial clinical evaluation [15]. The authors demonstrated that a model incorporating multiple ophthalmological and clinical variables could accurately classify and predict the progression of TAO. In a multicenter European study, a predictive scoring system was developed to estimate the risk of GO development or progression in newly diagnosed Graves patients [16]. This model primarily highlighted smoking and TSH receptor antibodies as key determinants; however, it mainly focused on predicting disease onset risk rather than classifying the existing severity or phenotype of GO. In contrast, our study addresses a different clinical question by modeling the clinical spectrum and severity of the disease in patients who already have a diagnosis of GO. In another study, Wang et al. utilized a random forest algorithm based solely on clinical and laboratory parameters to classify GO severity according to the EUGOGO criteria and reported high AUC values [17]. In the study by Lee et al., machine learning approaches were also utilized to develop clinically applicable risk scores for disease severity and the muscle-predominant phenotype of Graves ophthalmopathy, based on quantitative imaging parameters and clinical data [18]. However, a methodological limitation of that study is that only variables showing statistically significant differences between groups in conventional statistical analyses (p < 0.05) were included in the modeling process. Furthermore, the incorporation of parameters such as the NOSPECS classification, which already reflects disease severity, and clinical markers directly indicating tissue predominance, such as diplopia and ocular deviation, into the scoring system represents the inclusion of outcome indicators rather than true predictive factors. For this reason, in our study, we deliberately excluded direct physical examination findings from the predictive modeling and instead focused exclusively on demographic characteristics and biochemical parameters that have genuine predictive potential for clinical outcomes. In the machine learning analyses, the highest discriminative performance for predicting ophthalmopathy severity was achieved with the SVC, suggesting that complex and non-linear relationships within clinical data can be successfully captured. In the feature importance analysis, smoking emerged as the most influential variable for the severity outcome, clearly outweighing demographic factors such as age and sex, as well as biochemical parameters including thyroid hormone levels and autoantibodies. Although age, sex, and thyroid-related biomarkers are known to be associated with the presence and activity of Graves ophthalmopathy, within a multivariable predictive framework their effects may partially overlap with smoking or provide limited additional predictive value. In the NOSPECS scoring system, which represents another method of assessing disease severity, age emerged as the most influential factor, whereas smoking did not appear as prominently as in the EUGOGO-based evaluation. Since NOSPECS and EUGOGO reflect different clinical and biological dimensions of Graves ophthalmopathy, it is expected that feature importance outcomes derived from ML models based on these two scoring systems may differ. One of the distinguishing aspects of this study is the systematic evaluation of ML-based models for predicting clinical outcomes in GO. While conventional statistical methods are effective in assessing predefined linear relationships, their ability to fully characterize complex clinical scenarios involving multiple clinical, demographic, and biochemical variables is limited. ML algorithms, on the other hand, have the potential to capture nonlinear associations and intricate interactions among variables. Early identification of patients at higher risk of a more aggressive disease course using simple baseline data may facilitate closer monitoring and timely therapeutic intervention. This study has several limitations. Its retrospective design, single-center patient population, and heterogeneity in treatment approaches may limit the generalizability of the findings. In addition, the semi-quantitative nature of radiological assessments and the inability to analyze temporal changes in certain biochemical parameters represent further limitations. Nevertheless, the relatively large sample size, use of standardized classification systems, and combined evaluation of both conventional statistical methods and ML approaches constitute important strengths of this study. In this study, the clinical course and severity of Graves ophthalmopathy were comprehensively evaluated using both conventional statistical methods and ML models. Our findings demonstrate that demographic and clinical factors are more influential than thyroid hormone levels and autoantibody titers in predicting disease severity and activity. Furthermore, ML-based approaches were shown to provide meaningful discriminative performance in forecasting complex clinical outcomes. What is known before: Graves ophthalmopathy shows wide variation in severity, activity, onset pattern and tissue involvement, and predicting clinical course remains challenging. Smoking, sex and age are recognised as key risk factors, but their clinical relevance and predictive strength have not been fully established in routine practice. What this study adds: Demographic and clinical characteristics are shown to be more informative than routine thyroid hormone levels and autoantibody titres in characterising disease severity, activity and phenotype. It demonstrates that machine learning approaches can meaningfully support clinical risk assessment by identifying key predictors of disease severity, activity and phenotype. The findings support the potential integration of AI-based models alongside conventional assessment to enhance clinical decision-making in Graves ophthalmopathy. Declarations The requirement for informed consent was waived by the Kartal Dr.Lütfi Kırdar City Hospital Ethics Committee due to the retrospective nature of this case series. Conflict of interest: The authors declare that they have no competing interests. Author contributions: Concept and design: AA,TY,NG Data collection: ACÖ,AA Data analysis and interpretation: MO, FÇ Statistical analysis / Machine learning analysis: UD, AA Drafting of the manuscript: AA, FÇ Funding: None Acknowledgements: None References Chin YH, Ng CH, Lee MH, Koh JWH, Kiew J, Yang SP, et al. Prevalence of thyroid eye disease in Graves’ disease: A meta-analysis and systematic review. Clinical Endocrinology. 2020;93(4):363–74. Bartalena L, Piantanida E, Gallo D, Lai A, Tanda ML. Epidemiology, Natural History, Risk Factors, and Prevention of Graves’ Orbitopathy. Front Endocrinol. 2020 Nov 30;11. Bartalena L, Kahaly GJ, Baldeschi L, Dayan CM, Eckstein A, Marcocci C, et al. The 2021 European Group on Graves’ orbitopathy (EUGOGO) clinical practice guidelines for the medical management of Graves’ orbitopathy. Eur J Endocrinol. 2021 Oct 1;185(4):G43–67. Cao J, Su Y, Chen Z, Ma C, Xiong W. The risk factors for Graves’ ophthalmopathy. Graefes Arch Clin Exp Ophthalmol. 2022 Apr 1;260(4):1043–54. Li Q, Ye H, Ding Y, Chen G, Liu Z, Xu J, et al. Clinical characteristics of moderate-to-severe thyroid associated ophthalmopathy in 354 Chinese cases. PLOS ONE. 2017 May 4;12(5):e0176064. Stan MN, Bahn RS. Risk Factors for Development or Deterioration of Graves’ Ophthalmopathy. Thyroid®. 2010 Jul;20(7):777–83. Oeverhaus M, Winkler L, Stähr K, Daser A, Bechrakis N, Stöhr M, et al. Influence of biological sex, age and smoking on Graves’ orbitopathy – a ten-year tertiary referral center analysis. Front Endocrinol. 2023 Apr 4;14. Wiersinga WM, Eckstein AK, Žarković M. Thyroid eye disease (Graves’ orbitopathy): clinical presentation, epidemiology, pathogenesis, and management. The Lancet Diabetes & Endocrinology. 2025 Jul 1;13(7):600–14. Bartalena L, Martino E, Marcocci C, Bogazzi F, Panicucci M, Velluzzi F, et al. More on smoking habits and Graves’ ophthalmopathy. J Endocrinol Invest. 1989 Nov 1;12(10):733–7. Kau HC, Wu SB, Tsai CC, Liu CJL, Wei YH. Cigarette Smoke Extract-Induced Oxidative Stress and Fibrosis-Related Genes Expression in Orbital Fibroblasts from Patients with Graves’ Ophthalmopathy. Oxidative Medicine and Cellular Longevity. 2016;2016(1):4676289. Görtz GE, Horstmann M, Aniol B, Reyes BD, Fandrey J, Eckstein A, et al. Hypoxia-Dependent HIF-1 Activation Impacts on Tissue Remodeling in Graves’ Ophthalmopathy—Implications for Smoking. J Clin Endocrinol Metab. 2016 Dec 1;101(12):4834–42. Wiersinga WM, Regensburg NI, Mourits MP. Differential Involvement of Orbital Fat and Extraocular Muscles in Graves’ Ophthalmopathy. European Thyroid Journal. 2013 Mar 1;2(1):14–21. Le Moli R, Pluchino A, Muscia V, Regalbuto C, Luciani B, Squatrito S, et al. Graves’ orbitopathy: extraocular muscle/total orbit area ratio is positively related to the Clinical Activity Score. Eur J Ophthalmol. 2012;22(3):301–8. Zhang X. Global research on artificial intelligence in thyroid-associated ophthalmopathy: A bibliometric analysis. Advances in Ophthalmology Practice and Research. 2024 Feb 1;4(1):1–7. Salvi M, Dazzi D, Pellistri I, Neri F, Wall JR. Classification and prediction of the progression of thyroid-associated ophthalmopathy by an artificial neural network. Ophthalmology. 2002 Sep 1;109(9):1703–8. Wiersinga W, Žarković M, Bartalena L, Donati S, Perros P, Okosieme O, et al. Predictive score for the development or progression of Graves’ orbitopathy in patients with newly diagnosed Graves’ hyperthyroidism. Eur J Endocrinol. 2018 Jun 1;178(6):635–43. Wang M, Li G, Dong L, Hou Z, Zhang J, Li D. Severity Identification of Graves Orbitopathy via Random Forest Algorithm. Horm Metab Res. 2024 Oct;56(10):706–11. Lee S, Yu J, Kim Y, Kim M, Lew H. Application of an Interpretable Machine Learning for Estimating Severity of Graves’ Orbitopathy Based on Initial Finding. Journal of Clinical Medicine. 2023 Jan;12(7):2640. Tables Table1. Statistical analysis of cases according to severity and activity. Ophthalmopathy Severity Disease Activity Mild (n=107) Moderate-Severe (n=46) P value Inactive (n=92) Active (n=61) P value Age 46,36 ± 12,01 49,97 ± 12,85 0,097 48 (38-57) 49 (41-56) 0,386 Gender Male Female 25 (%23,4) 82 (%76,6) 20 (%43,5) 26 (%56,5) 0,012 20 (%21,7) 72 (%78,3) 25 (%41) 36 (%59) 0,011 Graves disease duration 8 (5-10) 6,5 (5-9) 0,282 7,5 (4,5-10,5) 7 (5-10) 0,645 Ophthalmopathy duration 6 (4-8) 5,5 (4-8) 0,667 6 (3-8) 6 (4-8) 0,328 Duration between graves disease and ophthalmopathy 0 (0-1) 0 (0-1) 0,868 0 (0-2) 0 (0-1) 0,055 NOSPECS score 3 (2-4) 6 (5-7) <0,001 3 (2-4) 5 (4-6) <0,001 Clinical activiy score Inactive Active 76 (%71) 31 (%29) 16 (%34,8) 30 (%65,2) <0,001 Ophthamopathy disease severity Mild Moderate-Severe 76 (%82,6) 16 (%17,4) 31 (%50,8) 30 (%49,2) <0,001 Thyroid gland function Hyperthyroid Euthyroid Hypothyroid 78 (%72,9) 20 (%18,7) 9 (%8,4) 39 (%84,8) 4 (%8,7) 3 (%6,5) 0,248 68 (%73,9) 17 (%18,5) 7 (%7,6) 49 (%80,3) 7 (%11,5) 5 (%8,2) 0,507 Thyroid hormone levels FT3 FT4 TSH 5,54 (4,49-9,21) 12,78 (8,3-24,93) 0,07 (0,002-1,44) 5,6 (4,61-7,77) 12,28 (9,45-18,85) 0,29 (0,003-1,56) 0,858 0,711 0,575 5,56 (4,33-8,96) 12,4 (4,81-21,73) 0,12 (0,003-1,53) 5,55 (4,65-7,99) 13,03 (9,96-24,62) 0,03 (0,003-1,45) 0,602 0,240 0,228 Thyroid otoimmun antibody levels TSHR-AB ANTI-TPO 4,89 (1,8-13,73) 32,9 (6,25-133,5) 5,63 (1,86-16,7) 22,3 (3,8-78,6) 0,381 0,385 4,89 (1,79-16,25) 32,55 (7,8-128,75) 5,54 (2,06-13,25) 26,4 (3,2-73,7) 0,698 0,332 Smoking Yes No 17 (%15,9) 90 (%84,1) 24 (%52,2) 22 (%47,8) <0,001 21 (%22,8) 71 (%77,2) 20 (%32,8) 41 (%67,2) 0,173 Dry eye 30 (%28) 12 (%26,1) 0,804 31 (%33,7) 11 (%18) 0,034 Glaucoma 5 (%4,7) 6 (%13) 0,066 6 (%6,5) 5 (%8,2) 0,695 Hypertension 30 (%28) 10 (%21,7) 0,416 24 (%26,1) 16 (%26,2) 0,984 Diabetes mellitus 9 (%8,4) 5 (%10,9) 0,629 7 (%7,6) 7 (%11,5) 0,417 Additional otoimmun disorder 5 (%4,7) 3 (%6,5) 0,638 4 (%4,3) 4 (%6,6) 0,548 Lid retraction 66 (%61,7) 37 (%80,4) 0,023 54 (%58,7) 49 (%80,3) 0,005 Diplopia 7 (%6,5) 15 (%32,6) <0,001 8 (%8,7) 14 (%23) 0,014 Restricted eye movement 26 (%24,3) 36 (%78,3) <0,001 25 (%27,2) 37 (%60,7) <0,001 Propitozis 61 (%57) 41 (%89,1) <0,001 55 (%59,8) 47 (%77) 0,027 Hertel 20 (18-21,5) 22 (21-24) <0,001 20 (18-22) 21 (19-23) 0,026 Extraocular muscle involvement None One muscle Two muscles Three or more muscles 73 (%68,2) 15 (%14) 6 (%5,6) 13 (%12,1) 9 (%19,6) 9 (%19,6) 9 (%19,6) 19 (%41,3) <0,001 64 (%69,6) 12 (%12) 7 (%7,6) 10 (%10,9) 18 (%29,5) 13 (%21,3) 8 (%13,1) 22 (%36,1) <0,001 Thyroidectomy 40 (%37,4) 24 (%52,2) 0,089 34 (%37) 30 (%49,2) 0,133 Table2. Statistical analysis of cases according to ophthalmopathy onset and tissue dominancy. Ophthalmopathy Onset Time Tissue Dominancy Early onset (n=89) Late onset (n=64) P value Fat predominant (n=105) Muscle predominant (n=48) P value Age 49 (39-55) 48,5 (38,5-57,5) 0,891 44 (38-53) 52 (42,5-60) 0,012 Gender Male Female 30 (%33,7) 59 (%66,3) 15 (%23,4) 49 (%76,6) 0,169 22 (%21) 83 (%79) 23 (%47,9) 25 (%52,1) 0,001 Graves disease duration 6 (4-8) 9 (6-11) <0,001 8 (5-10) 6,5 (4-8) 0,040 Ophthalmopathy duration 6 (4-8) 6 (3-8) 0,210 6 (4-9) 5 (4-8) 0,388 Duration between graves disease and ophthalmopathy 0 (0-0) 2 (1-4) <0,001 0 (0-1) 0 (0-1) 0,139 NOSPECS score 4 (3-5) 4 (2-5) 0,054 4 (2-5) 5 (3-6) 0,001 Ophthamopathy disease severity Mild Moderate-Severe 62 (%69,7) 27 (%30,3) 45 (%70,3) 19 (%29,7) 0,931 84 (%80) 21 (%20) 23 (%47,9) 25 (%52,1) <0,001 Clinical activiy score Inactive Active 48 (%53,9) 41 (%46,1) 44 (%68,8) 20 (%31,2) 0,065 72 (%68,6) 33 (%31,4) 20 (%41,7) 28 (%58,3) 0,002 Thyroid gland function Hyperthyroid Euthyroid Hypothyroid 71 (%79,8) 11 (%12,4) 7 (%7,9) 46 (%71,9) 13 (%20,3) 5 (%7,8) 0,406 81 (%77,1) 19 (%18,1) 5 (%4,8) 36 (%75) 5 (%10,4) 7 (%14,6) 0,070 Thyroid hormone levels FT3 FT4 TSH 5,33 (4,34-7,99) 12,3 (8,24-16,79) 0,2 (0,002-1,81) 5,71 (4,81-10,07) 14,27 (9,53-30,15) 0,04 (0,003-0,79) 0,079 0,106 0,194 5,58 (4,52-8,97) 13,18 (9,51-23,72) 0,05 (0,002-1,36) 5,52 (4,49-8,5) 10,55 (4,25-22,12) 0,65 (0,008-1,81) 0,970 0,124 0,067 Thyroid otoimmun antibody levels TSHR-AB ANTI-TPO 4,68 (1,8-12,5) 30,9 (5,5-121) 6,39 (2,3-19,29) 31,7 (7,65-121,5) 0,233 0,510 4,65 (1,81-14,56) 30,9 (6,5-132,5) 7,22 (1,96-15,51) 30,1 (4,65-71,1) 0,495 0,624 Smoking Yes No 22 (%24,7) 67 (%75,3) 19 (%29,7) 45 (%70,3) 0,494 21 (%20) 84 (%80) 20 (%41,7) 28 (%58,3) 0,005 Dry eye 26 (%29,2) 16 (%25) 0,565 30 (%28,6) 12 (%25) 0,646 Glaucoma 7 (%7,9) 4 (%6,3) 0,703 5 (%4,8) 6 (%12,5) 0,086 Hypertension 24 (%27) 16 (%25) 0,785 30 (%28,6) 10 (%20,8) 0,312 Diabetes mellitus 9 (%10,1) 5 (%7,8) 0,626 12 (%11,4) 2 (%4,2) 0,148 Additional otoimmun disorder 4 (%4,5) 4 (%6,3) 0,630 5 (%4,8) 3 (%6,3) 0,701 Lid retraction 63 (%70,8) 40 (%62,5) 0,284 60 (%57,1) 43 (%89,6) <0,001 Diplopia 15 (%16,9) 7 (%10,9) 0,304 5 (%4,8) 17 (%35,4) <0,001 Restricted eye movement 41 (%46,1) 21 (%32,8) 0,099 22 (%21) 40 (%83,3) <0,001 Propitozis 61 (%68,5) 41 (%64,1) 0,562 68 (%64,8) 34 (%70,8) 0,460 Hertel 21 (19-22) 20 (18-22) 0,419 20 (18-22) 21 (19-22) 0,157 Extraocular muscle involvement None One muscle Two muscles Three or more muscles 44 (%49,4) 14 (%15,7) 9 (%10,1) 22 (%24,7) 38 (%59,4) 10 (%15,6) 6 (%9,4) 10 (%15,6) 0,536 82 (%78,1) 18 (%17,1) 5 (%4,8) 0 (%0) 0 (%0) 6 (%12,5) 10 (%20,8) 32 (%66,7) <0,001 Thyroidectomy 34 (%38,2) 30 (%46,9) 0,283 42 (%40) 22 (%45,8) 0,497 Additional Declarations There is no conflict of interest Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8484257","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":570224241,"identity":"93b419fa-2002-4968-bd93-b495a49a71e3","order_by":0,"name":"Anıl Ağaçkesen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAUlEQVRIiWNgGAWjYFACxsYDIEoCzKkAYmbmBkJaGpC0nAFpYSSkhYEBoYWxDWIIXuX8s5sbDvz4ZScv2d7+8HHhvNpo/naglh8V23BqkbhzsOFgb1+y4WyeM8bGM7cdz51xmLGBsefMbdzW3EhsOMDbw5wgJ5HDJs277VhuA1ALM2Mbbi3yQC0H//bUJ8jJP3/+m3fOsdz5hLQYALUc5vlxOEFagsGMmbehJncDIS2GIC2yDccNZ/bkGEvzHDuQuxGo5SA+v8jdSH/48M2fanmJ48cffuapqcudd/7wwQc/KvB4HwQg0QEGh8HkAfzqQeAPnFVHWPEoGAWjYBSMOAAAbNtihTRRVZ4AAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0003-4694-4683","institution":"Kartal Dr.Lütfi Kırdar City Hospital","correspondingAuthor":true,"prefix":"","firstName":"Anıl","middleName":"","lastName":"Ağaçkesen","suffix":""},{"id":570224242,"identity":"a5740a40-b4cf-45cb-8ddd-75945fb43bd7","order_by":1,"name":"Titap Yazicioglu","email":"","orcid":"","institution":"Kartal Lütfi Kırdar Education and tTraining Hospital","correspondingAuthor":false,"prefix":"","firstName":"Titap","middleName":"","lastName":"Yazicioglu","suffix":""},{"id":570224243,"identity":"0f634db3-09b1-4218-bf4d-103ec6e60c0c","order_by":2,"name":"Murat Oklar","email":"","orcid":"https://orcid.org/0000-0001-5254-0613","institution":"University of Health Sciences Türkiye","correspondingAuthor":false,"prefix":"","firstName":"Murat","middleName":"","lastName":"Oklar","suffix":""},{"id":570224244,"identity":"e895bc82-13d5-4321-9a6d-57d7bad9a4d4","order_by":3,"name":"Ahmet Can Özdemir","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Ahmet","middleName":"Can","lastName":"Özdemir","suffix":""},{"id":570224245,"identity":"25ea16ee-7c1f-49e4-a894-ad1f3e93b2d0","order_by":4,"name":"Fatih Çoban","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Fatih","middleName":"","lastName":"Çoban","suffix":""},{"id":570224246,"identity":"b13dde06-16ae-460f-ae23-eddd036dfa74","order_by":5,"name":"Naile Gökkaya","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Naile","middleName":"","lastName":"Gökkaya","suffix":""},{"id":570224247,"identity":"2c76325f-13cd-4443-98e0-f8ade8b6ff33","order_by":6,"name":"Uğur Dinç","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Uğur","middleName":"","lastName":"Dinç","suffix":""}],"badges":[],"createdAt":"2025-12-30 21:10:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8484257/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8484257/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":99813282,"identity":"b3c9d866-af6d-47be-9715-6eee2b3a3fd7","added_by":"auto","created_at":"2026-01-08 14:38:48","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":138429,"visible":true,"origin":"","legend":"","description":"","filename":"Image2.docx","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/e92ade23aecd421ea4ef25b1.docx"},{"id":99813270,"identity":"69c9f5cf-7ef7-4054-b51b-07f5ccbd06fd","added_by":"auto","created_at":"2026-01-08 14:38:46","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55047,"visible":true,"origin":"","legend":"","description":"","filename":"eyejournalml.docx","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/83b10826d3c1745ac0b6e310.docx"},{"id":99813232,"identity":"17ae840c-28f4-4519-9fbb-eff13710da36","added_by":"auto","created_at":"2026-01-08 14:38:41","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20137,"visible":true,"origin":"","legend":"","description":"","filename":"Table1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/94fd899b173efbce202a9789.docx"},{"id":100356208,"identity":"9bcf50d7-775f-4da8-869e-7832760d5181","added_by":"auto","created_at":"2026-01-16 06:57:33","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":233973,"visible":true,"origin":"","legend":"","description":"","filename":"Image3.docx","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/4f0d4ec518333c898b6e0ad3.docx"},{"id":99813244,"identity":"79bf774e-49a7-4abf-a319-6824c65be88e","added_by":"auto","created_at":"2026-01-08 14:38:44","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20254,"visible":true,"origin":"","legend":"","description":"","filename":"Table2.docx","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/93bbfc9dbd3716c96507252a.docx"},{"id":99813318,"identity":"3adab0c3-a173-4e74-ae86-c76a47668854","added_by":"auto","created_at":"2026-01-08 14:38:49","extension":"json","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8844,"visible":true,"origin":"","legend":"","description":"","filename":"EYE253790.json","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/f8a73226d1bc61caa77b7fc4.json"},{"id":99813363,"identity":"8a39706c-2a10-40ba-abff-a272f43def71","added_by":"auto","created_at":"2026-01-08 14:38:56","extension":"xml","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":103168,"visible":true,"origin":"","legend":"","description":"","filename":"EYE2537900enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/26a996a106d8e37b9ca68a99.xml"},{"id":99813327,"identity":"4f49112b-0662-4765-ba45-b72cedbae4b0","added_by":"auto","created_at":"2026-01-08 14:38:50","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":122785,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/0bbe30225c0705da2b11d1b6.png"},{"id":99813396,"identity":"4d5c89f7-4845-43cf-a204-62eb4efffe92","added_by":"auto","created_at":"2026-01-08 14:39:01","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":218310,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/41d51133d52dd61e4bf88750.png"},{"id":99813493,"identity":"dad8e6b0-8f8a-4b03-a1a3-53cc97865afa","added_by":"auto","created_at":"2026-01-08 14:39:12","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":33159,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/5c1cca7dc89b798c3cd97df2.png"},{"id":99813316,"identity":"ec892b6e-4532-47ec-ad2e-7059524253d5","added_by":"auto","created_at":"2026-01-08 14:38:49","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":50594,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/f4ee8c9dd8a9a56f5db40eee.png"},{"id":99813285,"identity":"b813cd74-8c91-4143-8d1b-8ce563ad59ba","added_by":"auto","created_at":"2026-01-08 14:38:48","extension":"xml","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":101155,"visible":true,"origin":"","legend":"","description":"","filename":"EYE2537900structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/c8a337c556d82077bf48bdad.xml"},{"id":99813269,"identity":"d2df590c-9eed-4a8c-895b-18e0c4bedb44","added_by":"auto","created_at":"2026-01-08 14:38:46","extension":"html","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":109127,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/74e67289c7e6a0dfd5027f60.html"},{"id":99813324,"identity":"4d1ce342-da62-4fe9-8257-22ca7a1e0289","added_by":"auto","created_at":"2026-01-08 14:38:50","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":122785,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves according to machine learning models of clinical outcomes\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/f07ed6aafc068390663527d7.png"},{"id":99813322,"identity":"adc1cc79-7a43-4fd4-9286-1a5b535756ac","added_by":"auto","created_at":"2026-01-08 14:38:50","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":218310,"visible":true,"origin":"","legend":"\u003cp\u003eFeature importance according to machine learning models of clinical outcomes\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/0fa41fa894890398009f1a26.png"},{"id":100376313,"identity":"1e8153f5-6cf1-45a0-9b9d-b7b6f75c9d0d","added_by":"auto","created_at":"2026-01-16 08:44:35","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1135580,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8484257/v1/326d986f-e390-4085-b924-c5881c4558ad.pdf"}],"financialInterests":"There is no conflict of interest","formattedTitle":"Exploring Clinical Determinants and Machine Learning–Based Prediction of Disease Outcomes in Graves Ophthalmopathy","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eGraves ophthalmopathy(GO) is the most common extrathyroidal manifestation of Graves disease, affecting the ocular adnexa and the orbit. Although it is recognized as an autoimmune inflammatory disorder, its underlying pathophysiological mechanisms have not been fully elucidated. Thyrotropin receptor antibodies are considered to play a central role in the pathogenesis; however, insulin-like growth factor-1 (IGF-1) receptors, environmental influences, and genetic factors also contribute to disease development. Clinically, the disease may present with eyelid retraction, proptosis, and diplopia, and in more severe cases may lead to sight-threatening complications such as corneal exposure and optic neuropathy.\u003c/p\u003e \u003cp\u003eOphthalmopathy occurs in approximately 40% of patients with Graves disease [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. While conservative management and control of risk factors are usually sufficient in mild cases, immunosuppressive therapy, radiotherapy, or surgical intervention may be required in severe disease. Smoking, elevated thyrotropin receptor antibody levels, thyroid dysfunction, radioactive iodine therapy, and hypercholesterolemia have been reported as important risk factors [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. In addition, selenium deficiency, female sex, and genetic predisposition have also been associated with increased risk [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. A large-scale study conducted in a Chinese population reported that older age and male sex were associated with more severe disease [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e].\u003c/p\u003e \u003cp\u003ePredicting whether ophthalmopathy will develop in patients with Graves disease, when it will occur, and how severe it will be remains challenging. Considerable interindividual variability in clinical presentation and disease course is thought to reflect the multifactorial and complex nature of the underlying pathophysiology. Machine learning (ML) techniques have the potential to identify complex patterns that may not be captured by conventional statistical methods and can model nonlinear interactions among multiple clinical, biochemical, and demographic variables. Integration of ML-based approaches into clinical practice may facilitate more accurate early risk stratification and prediction of disease prognosis. The aim of this study was to investigate risk factors associated with the development of ophthalmopathy in patients presenting to the ophthalmology outpatient clinic using both conventional statistical analyses and ML-based methods, and to contribute to the literature, particularly in terms of prognostic prediction.\u003c/p\u003e"},{"header":"MATERIALS AND METHODS","content":"\u003cp\u003e\u003cstrong\u003eStudy Design\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEthical approval for the study was obtained from the institutional ethics committee (Approval number: 2025/010.99/20/1). Patients diagnosed with Graves ophthalmopathy who presented to our oculoplastic clinic were retrospectively reviewed. Thyroid hormone levels (T3, T4, TSH) at the time of initial Graves disease diagnosis, thyroid functional status (hyperthyroid, euthyroid, or hypothyroid), thyroid autoantibodies (TSHR-Ab, Anti-TPO), age, and sex were recorded. In addition, the time interval between the diagnosis of Graves disease and the onset of ophthalmopathy symptoms was noted in years.\u003c/p\u003e\n\u003cp\u003eComprehensive ophthalmological examination findings were evaluated. In unilateral cases, the affected eye, and in bilateral cases, the eye with more severe involvement was included in the analysis. Upper eyelid retraction was defined as an eyelid position \u0026ge;1 mm above the limbus. Patients with constant diplopia were considered diplopia-positive, and those with marked limitation in any gaze position were considered to have motility restriction. Hertel exophthalmometry measurements were recorded, and values \u0026ge;21 mm or an interocular difference \u0026ge;2 mm were accepted as significant proptosis. Additional ophthalmic comorbidities such as dry eye and glaucoma, as well as systemic comorbidities including hypertension, diabetes mellitus, and other autoimmune diseases, were documented. History of thyroidectomy and current smoking status were also recorded.\u003c/p\u003e\n\u003cp\u003eDisease severity was classified according to the European Group ON Graves Orbitopathy (EUGOGO) 2021 criteria as mild or moderate\u0026ndash;severe; additionally, NOSPECS scores were calculated for all patients. Disease activity was assessed using the Clinical Activity Score (CAS) in accordance with the EUGOGO 2021 guidelines; CAS \u0026ge;3 was defined as active and \u0026lt;3 as inactive ophthalmopathy. Patients who developed ophthalmopathy simultaneously with Graves disease or within 6 months were classified as early-onset, whereas those who developed ophthalmopathy later were classified as late-onset.\u003c/p\u003e\n\u003cp\u003eOrbital magnetic resonance imaging (MRI) scans were retrospectively evaluated. Based on tissue predominance, patients were classified into muscle-predominant and fat-predominant subtypes, as described in the literature. Muscle-predominant ophthalmopathy was defined as prominent enlargement and thickening of the extraocular muscles with comparatively less orbital fat expansion, whereas fat-predominant ophthalmopathy was characterized by marked orbital fat expansion with minimal or mild extraocular muscle involvement. Evaluations were performed semi-quantitatively by experienced ophthalmologists, taking radiological findings into consideration.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMachine Learning Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll analyses were performed in Python 3.12 using the \u003cem\u003epandas\u003c/em\u003e and \u003cem\u003escikit-learn\u003c/em\u003e libraries. For continuous variables, missing values were imputed using the mean and subsequently standardized; for categorical variables, missing data were completed with the most frequent category followed by one-hot encoding. A column transformer was used to ensure that the same preprocessing procedures were consistently applied across all models. Each target variable (outcome) was analyzed independently using the full set of available predictor variables.\u003c/p\u003e\n\u003cp\u003eTo provide a model-independent evaluation framework, four different classifier families were compared:\u003c/p\u003e\n\u003col start=\"1\" type=\"1\"\u003e\n \u003cli\u003eLogistic Regression\u003c/li\u003e\n \u003cli\u003eRandom Forest Classifier\u003c/li\u003e\n \u003cli\u003eSupport Vector Classifier (SVC with radial basis function kernel)\u003c/li\u003e\n \u003cli\u003ek-Nearest Neighbors (k-NN)\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eHyperparameters for each model were optimized using a 30-iteration, stratified 5-fold randomized cross-validation search. To ensure full reproducibility, the random seed was fixed at 42. For binary outcomes, model optimization was guided by the area under the receiver operating characteristic curve (AUC), whereas the macro-averaged F1-score was used for multiclass endpoints.\u003c/p\u003e\n\u003cp\u003eThe dataset was randomly divided into training (80%) and test (20%) sets while preserving class balance. Following hyperparameter optimization, the best-performing model for each outcome was retrained on the full training dataset and subsequently evaluated on the independent test set. Model performance was summarized using accuracy, balanced accuracy, macro-F1 score, and AUC (for binary targets). ROC curves were plotted to visualize discriminative performance.\u003c/p\u003e\n\u003cp\u003eTo quantitatively assess the contribution of each predictor, permutation importance was calculated on the test set using 10 random permutations per feature. This model-agnostic approach quantifies the mean reduction in performance (AUC or F1-score) when the association between a given feature and the outcome is disrupted; therefore, higher importance values indicate stronger model dependency on that variable. To maintain comparability across different algorithmic families, permutation importance was preferred over internal model coefficient\u0026ndash;based measures, even for linear and tree-based models. Feature importance results were visualized using horizontal bar plots displaying mean importance values and corresponding standard deviations.\u003c/p\u003e\n\u003cp\u003eAll analyses were conducted using \u003cem\u003escikit-learn\u003c/em\u003e version 1.5, and visualizations were generated with \u003cem\u003ematplotlib\u003c/em\u003e. Randomness was fully synchronized across Python, NumPy, and \u003cem\u003escikit-learn\u003c/em\u003e to ensure deterministic results. All scripts were executed in a controlled virtual environment (Python venv) on a Windows 10 operating system.\u003c/p\u003e\n\u003cp\u003eStatistical analyses were performed using SPSS software (IBM SPSS Statistics, version 23.0). Normality of continuous variables was assessed separately for each group using the Shapiro\u0026ndash;Wilk test. Continuous variables with a normal distribution were presented as mean \u0026plusmn; standard deviation (Mean \u0026plusmn; SD), whereas non-normally distributed variables were expressed as median (Q1\u0026ndash;Q3). For comparisons between two groups, the Independent Samples t-test was used when normality assumptions were met, and the Mann\u0026ndash;Whitney U test was applied otherwise. Homogeneity of variances was evaluated using Levene\u0026rsquo;s test. Categorical variables were expressed as frequencies and percentages, and group comparisons were performed using the Pearson chi-square test or Fisher\u0026rsquo;s Exact test, when appropriate. A p-value \u0026lt; 0.05 was considered statistically significant.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003eMedical records of 153 patients (108 females, 45 males) were reviewed. The mean age of the patients was 47.45 \u0026plusmn; 12.34 years. Patients were classified into two groups in four separate analyses according to disease activity (active / inactive), severity (mild / moderate\u0026ndash;severe), onset time (early / late onset), and tissue involvement pattern (fat-predominant / muscle-predominant).\u003c/p\u003e\n\u003cp\u003eWhen classified according to disease severity, the moderate\u0026ndash;severe group demonstrated significantly higher clinical activity scores, a higher proportion of male sex, and a significantly greater rate of active smoking. When grouped according to disease activity, the active group had a significantly higher proportion of male sex and greater disease severity; however, no significant difference was observed in terms of smoking status. Additionally, the frequency of dry eye was significantly higher in the inactive group compared with the active group.\u003c/p\u003e\n\u003cp\u003eWith respect to clinical examination findings, eyelid retraction, diplopia, restricted ocular motility, proptosis, higher Hertel measurements, and extraocular muscle involvement were significantly more frequent both in the moderate\u0026ndash;severe severity group and in the active disease group. Statistical comparisons according to disease severity and activity are summarized in Table 1.\u003c/p\u003e\n\u003cp\u003eTable1. Statistical analysis of cases according to severity and activity.\u003c/p\u003e\n\u003cp\u003eWhen patients were evaluated according to the onset time of ophthalmopathy, those with late-onset disease had a significantly longer follow-up duration. However, no statistically significant difference was observed between early- and late-onset groups with respect to the evaluated clinical and demographic parameters (Table 2).\u003c/p\u003e\n\u003cp\u003eRegarding tissue involvement patterns, patients with muscle-predominant ophthalmopathy tended to be significantly older, more frequently male, and more likely to be active smokers. In addition, active disease and moderate\u0026ndash;severe severity were more commonly observed in the muscle-predominant group. Clinical examination findings such as eyelid retraction, diplopia, and restricted ocular motility were also significantly more frequent in patients with muscle-predominant involvement (Table 2).\u003c/p\u003e\n\u003cp\u003eAcross all analyses, no statistically significant differences were detected between the groups in terms of thyroid hormone levels, thyroid functional status, or thyroid autoantibodies (Tables 1 and 2).\u003c/p\u003e\n\u003cp\u003eTable2. Statistical analysis of cases according to ophthalmopathy onset and tissue dominancy.\u003c/p\u003e\n\u003cp\u003eFollowing the application of machine learning models, the best predictive performance for ophthalmopathy severity was achieved with the Support Vector Classifier (SVC), reaching an AUC of 0.818. For ophthalmopathy onset, the highest performance was obtained with logistic regression, whereas Random Forest yielded the best performance for tissue predominance with an AUC of 0.662. NOSPECS score and disease activity were best predicted by Random Forest and k-Nearest Neighbors (k-NN), both achieving an AUC of approximately 0.64. The predictive performance levels of the clinical outcomes across different models and the corresponding ROC curves are summarized in Figure1.\u003c/p\u003e\n\u003cp\u003eFeature importance analyses demonstrated that machine learning models highlighted different predictors for different clinical outcomes (Figure2). In the model classifying disease severity according to EUGOGO, smoking status emerged as the strongest predictor, whereas in the NOSPECS-based classification, age was the most influential variable, with other parameters contributing at relatively similar levels. In the model predicting disease activity, sex, thyroid hormone levels, and thyroid autoantibodies provided substantial predictive contribution. For distinguishing early- versus late-onset ophthalmopathy, age was identified as the most important determinant. In the tissue predominance model, sex and TSH receptor autoantibodies played a prominent predictive role.\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eThe etiopathogenesis of Graves ophthalmopathy has not yet been fully elucidated and the disease is considered multifactorial. In addition to genetic and environmental influences, factors such as smoking, sex, age, and thyroid autoantibodies are also thought to play a role in its development and pathogenesis [6]. In the present study, the clinical characteristics of ophthalmopathy in patients with Graves disease were comprehensively evaluated using both conventional statistical methods and ML-based models.\u003c/p\u003e\n\u003cp\u003eIn our study, evaluation of disease severity demonstrated that male sex and active smoking were significantly associated with more severe ophthalmopathy. Similarly, the proportion of male patients was significantly higher in the active disease group. These findings are consistent with results from previous large-scale studies\u0026nbsp;[7,8]. It has been suggested that a more pronounced inflammatory response and a tendency for delayed presentation to medical care in male patients may contribute to this observation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSmoking is considered one of the most important modifiable risk factors in the pathogenesis of GO\u0026nbsp;[2]. It is well established that GO develops more frequently in patients with Graves disease who smoke, and that the disease follows a more severe course in smokers with GO\u0026nbsp;[9]. Smoking has been reported to enhance oxidative stress and induce fibrosis-related gene expression in orbital fibroblasts\u0026nbsp;[10]. Another molecular study demonstrated that smoking promotes hypoxia, thereby triggering adipogenesis and angiogenesis within the orbit\u0026nbsp;[11]. In our study, the strong association between smoking and disease severity is consistent with these proposed pathophysiological mechanisms.\u003c/p\u003e\n\u003cp\u003eWhen clinical findings related to disease activity and severity were evaluated, eyelid retraction, diplopia, restricted ocular motility, proptosis, and increased Hertel measurements were significantly more frequent in both the active and moderate\u0026ndash;severe groups. Furthermore, an increased number of extraocular muscle involvements was strongly associated with both disease activity and severity, suggesting that orbital muscle involvement and mechanical restriction reflect more advanced stages of the disease and parallel functional deterioration.\u003c/p\u003e\n\u003cp\u003eIn the analysis based on tissue predominance, the muscle-predominant subtype was found to be associated with older age, male sex, and smoking, and it was accompanied by more active and more severe ophthalmopathy. Consistent with our findings, Wiersinga et al. reported greater extraocular muscle enlargement in older patients, active smokers, and those with more severe ophthalmopathy in a radiological study tissue\u0026nbsp;[12]. Differences in fibroblast phenotypes, cytokine profiles, and orbital T-cell subsets have been proposed as potential cellular and molecular mechanisms contributing to the differential involvement of fat and muscle tissue\u0026nbsp;[12]. In addition, studies comparing extraocular muscle volume with total orbital volume have demonstrated that increased muscle volume is associated with active orbitopathy\u0026nbsp;[13]. In our study group, the finding of higher disease severity in muscle-dominant patients suggests that this subtype represents a more clinically aggressive phenotype. Collectively, these findings indicate that muscle-predominant involvement may serve not only as a morphological subclassification but also as an important prognostic indicator.\u003c/p\u003e\n\u003cp\u003eAnother noteworthy finding of our study is that thyroid hormone levels, thyroid functional status, and autoantibody titers did not differ significantly among any of the clinical subgroups in terms of disease severity, activity, onset pattern, or tissue predominance. This observation supports the concept that Graves ophthalmopathy is not merely a direct reflection of systemic thyroid autoimmunity; rather, local orbital immune responses, environmental influences, and individual susceptibility may play a more dominant role. From a clinical perspective, this suggests that relying solely on biochemical parameters for prognostic prediction in Graves ophthalmopathy may be insufficient.\u003c/p\u003e\n\u003cp\u003eIn recent years, the use of artificial intelligence and machine learning (ML)-based models in the field of thyroid ophthalmopathy has markedly increased, particularly for disease prediction and risk stratification\u0026nbsp;[14]. One of the earliest examples of AI application in Graves ophthalmopathy was reported by Salvi et al. in 2002, where an artificial neural network was employed to predict disease progression at the time of initial clinical evaluation\u0026nbsp;[15]. The authors demonstrated that a model incorporating multiple ophthalmological and clinical variables could accurately classify and predict the progression of TAO. In a multicenter European study, a predictive scoring system was developed to estimate the risk of GO development or progression in newly diagnosed Graves patients\u0026nbsp;[16]. This model primarily highlighted smoking and TSH receptor antibodies as key determinants; however, it mainly focused on predicting disease onset risk rather than classifying the existing severity or phenotype of GO. In contrast, our study addresses a different clinical question by modeling the clinical spectrum and severity of the disease in patients who already have a diagnosis of GO. In another study, Wang et al. utilized a random forest algorithm based solely on clinical and laboratory parameters to classify GO severity according to the EUGOGO criteria and reported high AUC values\u0026nbsp;[17].\u003c/p\u003e\n\u003cp\u003eIn the study by Lee et al., machine learning approaches were also utilized to develop clinically applicable risk scores for disease severity and the muscle-predominant phenotype of Graves ophthalmopathy, based on quantitative imaging parameters and clinical data\u0026nbsp;[18]. However, a methodological limitation of that study is that only variables showing statistically significant differences between groups in conventional statistical analyses (p \u0026lt; 0.05) were included in the modeling process. Furthermore, the incorporation of parameters such as the NOSPECS classification, which already reflects disease severity, and clinical markers directly indicating tissue predominance, such as diplopia and ocular deviation, into the scoring system represents the inclusion of outcome indicators rather than true predictive factors. For this reason, in our study, we deliberately excluded direct physical examination findings from the predictive modeling and instead focused exclusively on demographic characteristics and biochemical parameters that have genuine predictive potential for clinical outcomes.\u003c/p\u003e\n\u003cp\u003eIn the machine learning analyses, the highest discriminative performance for predicting ophthalmopathy severity was achieved with the SVC, suggesting that complex and non-linear relationships within clinical data can be successfully captured. In the feature importance analysis, smoking emerged as the most influential variable for the severity outcome, clearly outweighing demographic factors such as age and sex, as well as biochemical parameters including thyroid hormone levels and autoantibodies. Although age, sex, and thyroid-related biomarkers are known to be associated with the presence and activity of Graves ophthalmopathy, within a multivariable predictive framework their effects may partially overlap with smoking or provide limited additional predictive value.\u003c/p\u003e\n\u003cp\u003eIn the NOSPECS scoring system, which represents another method of assessing disease severity, age emerged as the most influential factor, whereas smoking did not appear as prominently as in the EUGOGO-based evaluation. Since NOSPECS and EUGOGO reflect different clinical and biological dimensions of Graves ophthalmopathy, it is expected that feature importance outcomes derived from ML models based on these two scoring systems may differ.\u003c/p\u003e\n\u003cp\u003eOne of the distinguishing aspects of this study is the systematic evaluation of ML-based models for predicting clinical outcomes in GO. While conventional statistical methods are effective in assessing predefined linear relationships, their ability to fully characterize complex clinical scenarios involving multiple clinical, demographic, and biochemical variables is limited. ML algorithms, on the other hand, have the potential to capture nonlinear associations and intricate interactions among variables. Early identification of patients at higher risk of a more aggressive disease course using simple baseline data may facilitate closer monitoring and timely therapeutic intervention.\u003c/p\u003e\n\u003cp\u003eThis study has several limitations. Its retrospective design, single-center patient population, and heterogeneity in treatment approaches may limit the generalizability of the findings. In addition, the semi-quantitative nature of radiological assessments and the inability to analyze temporal changes in certain biochemical parameters represent further limitations. Nevertheless, the relatively large sample size, use of standardized classification systems, and combined evaluation of both conventional statistical methods and ML approaches constitute important strengths of this study.\u003c/p\u003e\n\u003cp\u003eIn this study, the clinical course and severity of Graves ophthalmopathy were comprehensively evaluated using both conventional statistical methods and ML models. Our findings demonstrate that demographic and clinical factors are more influential than thyroid hormone levels and autoantibody titers in predicting disease severity and activity. Furthermore, ML-based approaches were shown to provide meaningful discriminative performance in forecasting complex clinical outcomes. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWhat is known before:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eGraves ophthalmopathy shows wide variation in severity, activity, onset pattern and tissue involvement, and predicting clinical course remains challenging.\u003c/li\u003e\n \u003cli\u003eSmoking, sex and age are recognised as key risk factors, but their clinical relevance and predictive strength have not been fully established in routine practice.\u0026nbsp;\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eWhat this study adds:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eDemographic and clinical characteristics are shown to be more informative than routine thyroid hormone levels and autoantibody titres in characterising disease severity, activity and phenotype.\u003c/li\u003e\n \u003cli\u003eIt demonstrates that machine learning approaches can meaningfully support clinical risk assessment by identifying key predictors of disease severity, activity and phenotype.\u003c/li\u003e\n \u003cli\u003eThe findings support the potential integration of AI-based models alongside conventional assessment to enhance clinical decision-making in Graves ophthalmopathy.\u003c/li\u003e\n\u003c/ul\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cspan\u003eThe requirement for informed consent was waived by the Kartal Dr.L\u0026uuml;tfi Kırdar City Hospital Ethics Committee due to the retrospective nature of this case series.\u003c/span\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eConflict of interest:\u003c/strong\u003e The authors declare that they have no competing interests.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions:\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConcept and design: AA,TY,NG\u003cbr\u003e\u0026nbsp;Data collection: AC\u0026Ouml;,AA\u003cbr\u003e\u0026nbsp;Data analysis and interpretation: MO, F\u0026Ccedil;\u003cbr\u003e\u0026nbsp;Statistical analysis / Machine learning analysis: UD, AA\u003cbr\u003e\u0026nbsp;Drafting of the manuscript: AA, F\u0026Ccedil;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e None\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements:\u003c/strong\u003e None\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eChin YH, Ng CH, Lee MH, Koh JWH, Kiew J, Yang SP, et al. Prevalence of thyroid eye disease in Graves\u0026rsquo; disease: A meta-analysis and systematic review. Clinical Endocrinology. 2020;93(4):363\u0026ndash;74. \u003c/li\u003e\n\u003cli\u003eBartalena L, Piantanida E, Gallo D, Lai A, Tanda ML. Epidemiology, Natural History, Risk Factors, and Prevention of Graves\u0026rsquo; Orbitopathy. Front Endocrinol. 2020 Nov 30;11. \u003c/li\u003e\n\u003cli\u003eBartalena L, Kahaly GJ, Baldeschi L, Dayan CM, Eckstein A, Marcocci C, et al. The 2021 European Group on Graves\u0026rsquo; orbitopathy (EUGOGO) clinical practice guidelines for the medical management of Graves\u0026rsquo; orbitopathy. Eur J Endocrinol. 2021 Oct 1;185(4):G43\u0026ndash;67. \u003c/li\u003e\n\u003cli\u003eCao J, Su Y, Chen Z, Ma C, Xiong W. The risk factors for Graves\u0026rsquo; ophthalmopathy. Graefes Arch Clin Exp Ophthalmol. 2022 Apr 1;260(4):1043\u0026ndash;54. \u003c/li\u003e\n\u003cli\u003eLi Q, Ye H, Ding Y, Chen G, Liu Z, Xu J, et al. Clinical characteristics of moderate-to-severe thyroid associated ophthalmopathy in 354 Chinese cases. PLOS ONE. 2017 May 4;12(5):e0176064. \u003c/li\u003e\n\u003cli\u003eStan MN, Bahn RS. Risk Factors for Development or Deterioration of Graves\u0026rsquo; Ophthalmopathy. Thyroid\u0026reg;. 2010 Jul;20(7):777\u0026ndash;83. \u003c/li\u003e\n\u003cli\u003eOeverhaus M, Winkler L, St\u0026auml;hr K, Daser A, Bechrakis N, St\u0026ouml;hr M, et al. Influence of biological sex, age and smoking on Graves\u0026rsquo; orbitopathy \u0026ndash; a ten-year tertiary referral center analysis. Front Endocrinol. 2023 Apr 4;14. \u003c/li\u003e\n\u003cli\u003eWiersinga WM, Eckstein AK, Žarković M. Thyroid eye disease (Graves\u0026rsquo; orbitopathy): clinical presentation, epidemiology, pathogenesis, and management. The Lancet Diabetes \u0026amp; Endocrinology. 2025 Jul 1;13(7):600\u0026ndash;14. \u003c/li\u003e\n\u003cli\u003eBartalena L, Martino E, Marcocci C, Bogazzi F, Panicucci M, Velluzzi F, et al. More on smoking habits and Graves\u0026rsquo; ophthalmopathy. J Endocrinol Invest. 1989 Nov 1;12(10):733\u0026ndash;7. \u003c/li\u003e\n\u003cli\u003eKau HC, Wu SB, Tsai CC, Liu CJL, Wei YH. Cigarette Smoke Extract-Induced Oxidative Stress and Fibrosis-Related Genes Expression in Orbital Fibroblasts from Patients with Graves\u0026rsquo; Ophthalmopathy. Oxidative Medicine and Cellular Longevity. 2016;2016(1):4676289. \u003c/li\u003e\n\u003cli\u003eG\u0026ouml;rtz GE, Horstmann M, Aniol B, Reyes BD, Fandrey J, Eckstein A, et al. Hypoxia-Dependent HIF-1 Activation Impacts on Tissue Remodeling in Graves\u0026rsquo; Ophthalmopathy\u0026mdash;Implications for Smoking. J Clin Endocrinol Metab. 2016 Dec 1;101(12):4834\u0026ndash;42. \u003c/li\u003e\n\u003cli\u003eWiersinga WM, Regensburg NI, Mourits MP. Differential Involvement of Orbital Fat and Extraocular Muscles in Graves\u0026rsquo; Ophthalmopathy. European Thyroid Journal. 2013 Mar 1;2(1):14\u0026ndash;21. \u003c/li\u003e\n\u003cli\u003eLe Moli R, Pluchino A, Muscia V, Regalbuto C, Luciani B, Squatrito S, et al. Graves\u0026rsquo; orbitopathy: extraocular muscle/total orbit area ratio is positively related to the Clinical Activity Score. Eur J Ophthalmol. 2012;22(3):301\u0026ndash;8. \u003c/li\u003e\n\u003cli\u003eZhang X. Global research on artificial intelligence in thyroid-associated ophthalmopathy: A bibliometric analysis. Advances in Ophthalmology Practice and Research. 2024 Feb 1;4(1):1\u0026ndash;7. \u003c/li\u003e\n\u003cli\u003eSalvi M, Dazzi D, Pellistri I, Neri F, Wall JR. Classification and prediction of the progression of thyroid-associated ophthalmopathy by an artificial neural network. Ophthalmology. 2002 Sep 1;109(9):1703\u0026ndash;8. \u003c/li\u003e\n\u003cli\u003eWiersinga W, Žarković M, Bartalena L, Donati S, Perros P, Okosieme O, et al. Predictive score for the development or progression of Graves\u0026rsquo; orbitopathy in patients with newly diagnosed Graves\u0026rsquo; hyperthyroidism. Eur J Endocrinol. 2018 Jun 1;178(6):635\u0026ndash;43. \u003c/li\u003e\n\u003cli\u003eWang M, Li G, Dong L, Hou Z, Zhang J, Li D. Severity Identification of Graves Orbitopathy via Random Forest Algorithm. Horm Metab Res. 2024 Oct;56(10):706\u0026ndash;11. \u003c/li\u003e\n\u003cli\u003eLee S, Yu J, Kim Y, Kim M, Lew H. Application of an Interpretable Machine Learning for Estimating Severity of Graves\u0026rsquo; Orbitopathy Based on Initial Finding. Journal of Clinical Medicine. 2023 Jan;12(7):2640. \u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable1. Statistical analysis of cases according to severity and activity.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 236px;\"\u003e\n \u003cp\u003eOphthalmopathy Severity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 246px;\"\u003e\n \u003cp\u003eDisease Activity\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eMild\u003c/p\u003e\n \u003cp\u003e(n=107)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eModerate-Severe\u003c/p\u003e\n \u003cp\u003e(n=46)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003eP value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eInactive\u003c/p\u003e\n \u003cp\u003e(n=92)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eActive\u003c/p\u003e\n \u003cp\u003e(n=61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eP value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e46,36 \u0026plusmn; 12,01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e49,97 \u0026plusmn; 12,85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,097\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e48 (38-57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e49 (41-56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,386\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e25 (%23,4)\u003c/p\u003e\n \u003cp\u003e82 (%76,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e20 (%43,5)\u003c/p\u003e\n \u003cp\u003e26 (%56,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,012\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e20 (%21,7)\u003c/p\u003e\n \u003cp\u003e72 (%78,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e25 (%41)\u003c/p\u003e\n \u003cp\u003e36 (%59)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,011\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eGraves disease duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e8 (5-10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e6,5 (5-9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,282\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e7,5 (4,5-10,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e7 (5-10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,645\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eOphthalmopathy duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e6 (4-8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e5,5 (4-8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,667\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e6 (3-8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e6 (4-8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,328\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eDuration between graves disease and ophthalmopathy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e0 (0-1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e0 (0-1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,868\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e0 (0-2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e0 (0-1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,055\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eNOSPECS score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e3 (2-4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e6 (5-7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e3 (2-4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5 (4-6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eClinical activiy score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eInactive\u003c/p\u003e\n \u003cp\u003eActive\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e76 (%71)\u003c/p\u003e\n \u003cp\u003e31 (%29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e16 (%34,8)\u003c/p\u003e\n \u003cp\u003e30 (%65,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eOphthamopathy disease severity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eMild\u003c/p\u003e\n \u003cp\u003eModerate-Severe\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e76 (%82,6)\u003c/p\u003e\n \u003cp\u003e16 (%17,4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e31 (%50,8)\u003c/p\u003e\n \u003cp\u003e30 (%49,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eThyroid gland function\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eHyperthyroid\u003c/p\u003e\n \u003cp\u003eEuthyroid\u003c/p\u003e\n \u003cp\u003eHypothyroid\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e78 (%72,9)\u003c/p\u003e\n \u003cp\u003e20 (%18,7)\u003c/p\u003e\n \u003cp\u003e9 (%8,4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e39 (%84,8)\u003c/p\u003e\n \u003cp\u003e4 (%8,7)\u003c/p\u003e\n \u003cp\u003e3 (%6,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,248\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e68 (%73,9)\u003c/p\u003e\n \u003cp\u003e17 (%18,5)\u003c/p\u003e\n \u003cp\u003e7 (%7,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e49 (%80,3)\u003c/p\u003e\n \u003cp\u003e7 (%11,5)\u003c/p\u003e\n \u003cp\u003e5 (%8,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,507\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eThyroid hormone levels\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eFT3\u003c/p\u003e\n \u003cp\u003eFT4\u003c/p\u003e\n \u003cp\u003eTSH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5,54 (4,49-9,21)\u003c/p\u003e\n \u003cp\u003e12,78 (8,3-24,93)\u003c/p\u003e\n \u003cp\u003e0,07 (0,002-1,44)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e5,6 (4,61-7,77)\u003c/p\u003e\n \u003cp\u003e12,28 (9,45-18,85)\u003c/p\u003e\n \u003cp\u003e0,29 (0,003-1,56)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,858\u003c/p\u003e\n \u003cp\u003e0,711\u003c/p\u003e\n \u003cp\u003e0,575\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e5,56 (4,33-8,96)\u003c/p\u003e\n \u003cp\u003e12,4 (4,81-21,73)\u003c/p\u003e\n \u003cp\u003e0,12 (0,003-1,53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5,55 (4,65-7,99)\u003c/p\u003e\n \u003cp\u003e13,03 (9,96-24,62)\u003c/p\u003e\n \u003cp\u003e0,03 (0,003-1,45)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,602\u003c/p\u003e\n \u003cp\u003e0,240\u003c/p\u003e\n \u003cp\u003e0,228\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eThyroid otoimmun \u0026nbsp;antibody levels\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eTSHR-AB\u003c/p\u003e\n \u003cp\u003eANTI-TPO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e4,89 (1,8-13,73)\u003c/p\u003e\n \u003cp\u003e32,9 (6,25-133,5)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e5,63 (1,86-16,7)\u003c/p\u003e\n \u003cp\u003e22,3 (3,8-78,6)\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,381\u003c/p\u003e\n \u003cp\u003e0,385\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e4,89 (1,79-16,25)\u003c/p\u003e\n \u003cp\u003e32,55 (7,8-128,75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5,54 (2,06-13,25)\u003c/p\u003e\n \u003cp\u003e26,4 (3,2-73,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,698\u003c/p\u003e\n \u003cp\u003e0,332\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eSmoking\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e17 (%15,9)\u003c/p\u003e\n \u003cp\u003e90 (%84,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e24 (%52,2)\u003c/p\u003e\n \u003cp\u003e22 (%47,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e21 (%22,8)\u003c/p\u003e\n \u003cp\u003e71 (%77,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e20 (%32,8)\u003c/p\u003e\n \u003cp\u003e41 (%67,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,173\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eDry eye\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e30 (%28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e12 (%26,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,804\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e31 (%33,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e11 (%18)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,034\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eGlaucoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5 (%4,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e6 (%13)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,066\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e6 (%6,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5 (%8,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,695\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eHypertension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e30 (%28)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e10 (%21,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,416\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e24 (%26,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e16 (%26,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,984\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eDiabetes mellitus\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e9 (%8,4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e5 (%10,9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,629\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e7 (%7,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e7 (%11,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,417\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eAdditional otoimmun disorder\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5 (%4,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e3 (%6,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,638\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e4 (%4,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e4 (%6,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,548\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eLid retraction\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e66 (%61,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e37 (%80,4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,023\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e54 (%58,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e49 (%80,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,005\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eDiplopia\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e7 (%6,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e15 (%32,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e8 (%8,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e14 (%23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,014\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eRestricted eye movement\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e26 (%24,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e36 (%78,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e25 (%27,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e37 (%60,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003ePropitozis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e61 (%57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e41 (%89,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e55 (%59,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e47 (%77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,027\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eHertel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e20 (18-21,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e22 (21-24)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e20 (18-22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e21 (19-23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,026\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eExtraocular muscle involvement\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eNone\u003c/p\u003e\n \u003cp\u003eOne muscle\u003c/p\u003e\n \u003cp\u003eTwo muscles\u003c/p\u003e\n \u003cp\u003eThree or more muscles\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e73 (%68,2)\u003c/p\u003e\n \u003cp\u003e15 (%14)\u003c/p\u003e\n \u003cp\u003e6 (%5,6)\u003c/p\u003e\n \u003cp\u003e13 (%12,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e9 (%19,6)\u003c/p\u003e\n \u003cp\u003e9 (%19,6)\u003c/p\u003e\n \u003cp\u003e9 (%19,6)\u003c/p\u003e\n \u003cp\u003e19 (%41,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e64 (%69,6)\u003c/p\u003e\n \u003cp\u003e12 (%12)\u003c/p\u003e\n \u003cp\u003e7 (%7,6)\u003c/p\u003e\n \u003cp\u003e10 (%10,9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e18 (%29,5)\u003c/p\u003e\n \u003cp\u003e13 (%21,3)\u003c/p\u003e\n \u003cp\u003e8 (%13,1)\u003c/p\u003e\n \u003cp\u003e22 (%36,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eThyroidectomy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e40 (%37,4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e24 (%52,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,089\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e34 (%37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e30 (%49,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0,133\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable2. Statistical analysis of cases according to ophthalmopathy onset and tissue dominancy.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 243px;\"\u003e\n \u003cp\u003eOphthalmopathy Onset Time\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"top\" style=\"width: 239px;\"\u003e\n \u003cp\u003eTissue Dominancy\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eEarly onset\u003c/p\u003e\n \u003cp\u003e(n=89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eLate onset\u003c/p\u003e\n \u003cp\u003e(n=64)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003eP value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003eFat predominant\u003c/p\u003e\n \u003cp\u003e(n=105)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eMuscle predominant\u003c/p\u003e\n \u003cp\u003e(n=48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003eP value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e49 (39-55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e48,5 (38,5-57,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,891\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e44 (38-53)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e52 (42,5-60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,012\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e30 (%33,7)\u003c/p\u003e\n \u003cp\u003e59 (%66,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e15 (%23,4)\u003c/p\u003e\n \u003cp\u003e49 (%76,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,169\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e22 (%21)\u003c/p\u003e\n \u003cp\u003e83 (%79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e23 (%47,9)\u003c/p\u003e\n \u003cp\u003e25 (%52,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eGraves disease duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e6 (4-8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e9 (6-11)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e8 (5-10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e6,5 (4-8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,040\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eOphthalmopathy duration\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e6 (4-8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e6 (3-8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,210\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e6 (4-9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e5 (4-8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,388\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eDuration between graves disease and ophthalmopathy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e0 (0-0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e2 (1-4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e0 (0-1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e0 (0-1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,139\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eNOSPECS score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e4 (3-5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e4 (2-5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,054\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e4 (2-5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e5 (3-6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eOphthamopathy disease severity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eMild\u003c/p\u003e\n \u003cp\u003eModerate-Severe\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e62 (%69,7)\u003c/p\u003e\n \u003cp\u003e27 (%30,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e45 (%70,3)\u003c/p\u003e\n \u003cp\u003e19 (%29,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,931\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e84 (%80)\u003c/p\u003e\n \u003cp\u003e21 (%20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e23 (%47,9)\u003c/p\u003e\n \u003cp\u003e25 (%52,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eClinical activiy score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eInactive\u003c/p\u003e\n \u003cp\u003eActive\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e48 (%53,9)\u003c/p\u003e\n \u003cp\u003e41 (%46,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e44 (%68,8)\u003c/p\u003e\n \u003cp\u003e20 (%31,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,065\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e72 (%68,6)\u003c/p\u003e\n \u003cp\u003e33 (%31,4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e20 (%41,7)\u003c/p\u003e\n \u003cp\u003e28 (%58,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,002\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eThyroid gland function\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eHyperthyroid\u003c/p\u003e\n \u003cp\u003eEuthyroid\u003c/p\u003e\n \u003cp\u003eHypothyroid\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e71 (%79,8)\u003c/p\u003e\n \u003cp\u003e11 (%12,4)\u003c/p\u003e\n \u003cp\u003e7 (%7,9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e46 (%71,9)\u003c/p\u003e\n \u003cp\u003e13 (%20,3)\u003c/p\u003e\n \u003cp\u003e5 (%7,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,406\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e81 (%77,1)\u003c/p\u003e\n \u003cp\u003e19 (%18,1)\u003c/p\u003e\n \u003cp\u003e5 (%4,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e36 (%75)\u003c/p\u003e\n \u003cp\u003e5 (%10,4)\u003c/p\u003e\n \u003cp\u003e7 (%14,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,070\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eThyroid hormone levels\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eFT3\u003c/p\u003e\n \u003cp\u003eFT4\u003c/p\u003e\n \u003cp\u003eTSH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e5,33 (4,34-7,99)\u003c/p\u003e\n \u003cp\u003e12,3 (8,24-16,79)\u003c/p\u003e\n \u003cp\u003e0,2 (0,002-1,81)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5,71 (4,81-10,07)\u003c/p\u003e\n \u003cp\u003e14,27 (9,53-30,15)\u003c/p\u003e\n \u003cp\u003e0,04 (0,003-0,79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,079\u003c/p\u003e\n \u003cp\u003e0,106\u003c/p\u003e\n \u003cp\u003e0,194\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e5,58 (4,52-8,97)\u003c/p\u003e\n \u003cp\u003e13,18 (9,51-23,72)\u003c/p\u003e\n \u003cp\u003e0,05 (0,002-1,36)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e5,52 (4,49-8,5)\u003c/p\u003e\n \u003cp\u003e10,55 (4,25-22,12)\u003c/p\u003e\n \u003cp\u003e0,65 (0,008-1,81)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,970\u003c/p\u003e\n \u003cp\u003e0,124\u003c/p\u003e\n \u003cp\u003e0,067\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eThyroid otoimmun \u0026nbsp;antibody levels\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eTSHR-AB\u003c/p\u003e\n \u003cp\u003eANTI-TPO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e4,68 (1,8-12,5)\u003c/p\u003e\n \u003cp\u003e30,9 (5,5-121)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e6,39 (2,3-19,29)\u003c/p\u003e\n \u003cp\u003e31,7 (7,65-121,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,233\u003c/p\u003e\n \u003cp\u003e0,510\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e4,65 (1,81-14,56)\u003c/p\u003e\n \u003cp\u003e30,9 (6,5-132,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e7,22 (1,96-15,51)\u003c/p\u003e\n \u003cp\u003e30,1 (4,65-71,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,495\u003c/p\u003e\n \u003cp\u003e0,624\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eSmoking\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e22 (%24,7)\u003c/p\u003e\n \u003cp\u003e67 (%75,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e19 (%29,7)\u003c/p\u003e\n \u003cp\u003e45 (%70,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,494\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e21 (%20)\u003c/p\u003e\n \u003cp\u003e84 (%80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e20 (%41,7)\u003c/p\u003e\n \u003cp\u003e28 (%58,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e0,005\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eDry eye\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e26 (%29,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e16 (%25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,565\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e30 (%28,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e12 (%25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,646\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eGlaucoma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e7 (%7,9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e4 (%6,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,703\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e5 (%4,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e6 (%12,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,086\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eHypertension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e24 (%27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e16 (%25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,785\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e30 (%28,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e10 (%20,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,312\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eDiabetes mellitus\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e9 (%10,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e5 (%7,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,626\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e12 (%11,4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e2 (%4,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,148\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eAdditional otoimmun disorder\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e4 (%4,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e4 (%6,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,630\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e5 (%4,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e3 (%6,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,701\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eLid retraction\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e63 (%70,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e40 (%62,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,284\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e60 (%57,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e43 (%89,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eDiplopia\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e15 (%16,9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e7 (%10,9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,304\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e5 (%4,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e17 (%35,4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eRestricted eye movement\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e41 (%46,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e21 (%32,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,099\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e22 (%21)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e40 (%83,3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003ePropitozis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e61 (%68,5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e41 (%64,1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,562\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e68 (%64,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e34 (%70,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,460\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eHertel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e21 (19-22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e20 (18-22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,419\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e20 (18-22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e21 (19-22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,157\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eExtraocular muscle involvement\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003eNone\u003c/p\u003e\n \u003cp\u003eOne muscle\u003c/p\u003e\n \u003cp\u003eTwo muscles\u003c/p\u003e\n \u003cp\u003eThree or more muscles\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e44 (%49,4)\u003c/p\u003e\n \u003cp\u003e14 (%15,7)\u003c/p\u003e\n \u003cp\u003e9 (%10,1)\u003c/p\u003e\n \u003cp\u003e22 (%24,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e38 (%59,4)\u003c/p\u003e\n \u003cp\u003e10 (%15,6)\u003c/p\u003e\n \u003cp\u003e6 (%9,4)\u003c/p\u003e\n \u003cp\u003e10 (%15,6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,536\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e82 (%78,1)\u003c/p\u003e\n \u003cp\u003e18 (%17,1)\u003c/p\u003e\n \u003cp\u003e5 (%4,8)\u003c/p\u003e\n \u003cp\u003e0 (%0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e0 (%0)\u003c/p\u003e\n \u003cp\u003e6 (%12,5)\u003c/p\u003e\n \u003cp\u003e10 (%20,8)\u003c/p\u003e\n \u003cp\u003e32 (%66,7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026lt;0,001\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eThyroidectomy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e34 (%38,2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e30 (%46,9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e0,283\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e42 (%40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e22 (%45,8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 47px;\"\u003e\n \u003cp\u003e0,497\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8484257/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8484257/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eGraves ophthalmopathy (GO) exhibits heterogeneous clinical behavior, and predicting disease severity and course remains challenging. This study aimed to evaluate clinical determinants of GO using conventional statistical analyses and machine learning (ML) approaches to improve prognostic prediction.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eMedical records of 153 patients with GO were retrospectively reviewed. Demographic characteristics, clinical findings, thyroid function, autoantibody levels, imaging features, and smoking status were analyzed. Patients were classified according to disease severity, activity, onset pattern, and tissue predominance. In addition to classical statistical tests, logistic regression, random forest, support vector classifier (SVC), and k-nearest neighbors (k-NN) algorithms were trained and evaluated using accuracy, F1-score, and AUC metrics.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eMale sex and active smoking were significantly associated with higher disease severity and activity. Muscle-predominant GO was associated with older age, male sex, smoking, higher severity, and higher activity. No significant differences were observed among clinical subgroups regarding thyroid hormone levels or autoantibody titers. Among ML models, the SVC showed the best performance for predicting severity (AUC\u0026thinsp;=\u0026thinsp;0.818); Random Forest best estimated tissue predominance (AUC\u0026thinsp;=\u0026thinsp;0.662), while logistic regression showed the highest accuracy for onset prediction. Smoking was the strongest predictor of EUGOGO-based severity, whereas age was most influential in NOSPECS-based assessments.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eDemographic and clinical factors, particularly smoking, sex, age, and muscle involvement, appear to be stronger determinants of GO course than thyroid-related biochemical parameters. Machine learning approaches demonstrated meaningful discriminatory capability for predicting complex clinical outcomes and may contribute to future risk stratification frameworks.\u003c/p\u003e","manuscriptTitle":"Exploring Clinical Determinants and Machine Learning–Based Prediction of Disease Outcomes in Graves Ophthalmopathy","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-08 14:02:31","doi":"10.21203/rs.3.rs-8484257/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"23340405-71c7-401e-aa48-7adb3200e2f9","owner":[],"postedDate":"January 8th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":60675710,"name":"Health sciences/Risk factors"},{"id":60675711,"name":"Health sciences/Medical research/Outcomes research"}],"tags":[],"updatedAt":"2026-01-12T12:01:52+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-08 14:02:31","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8484257","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8484257","identity":"rs-8484257","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}