Search

Background: Few studies on rheumatoid arthritis (RA) have generated machine learning models to predict biologic disease-modifying antirheumatic drugs (bDMARDs) responses; however, these studies included insufficient analysis on important features. Moreover, machine learning is yet to be used to predict bDMARD responses in ankylosing spondylitis (AS). Thus, in this study, machine learning was used to predict such responses in RA and AS patients., Methods: Data were retrieved from the Korean College of Rheumatology Biologics therapy (KOBIO) registry. The number of RA and AS patients in the training dataset were 625 and 611, respectively. We prepared independent test datasets that did not participate in any process of generating machine learning models. Baseline clinical characteristics were used as input features. Responders were defined as those who met the ACR 20% improvement response criteria (ACR20) and ASAS 20% improvement response criteria (ASAS20) in RA and AS, respectively, at the first follow-up. Multiple machine learning methods, including random forest (RF-method), were used to generate models to predict bDMARD responses, and we compared them with the logistic regression model., Results: The RF-method model had superior prediction performance to logistic regression model (accuracy: 0.726 [95% confidence interval (CI): 0.725-0.730] vs. 0.689 [0.606-0.717], area under curve (AUC) of the receiver operating characteristic curve (ROC) 0.638 [0.576-0.658] vs. 0.565 [0.493-0.605], F1 score 0.841 [0.837-0.843] vs. 0.803 [0.732-0.828], AUC of the precision-recall curve 0.808 [0.763-0.829] vs. 0.754 [0.714-0.789]) with independent test datasets in patients with RA. However, machine learning and logistic regression exhibited similar prediction performance in AS patients. Furthermore, the patient self-reporting scales, which are patient global assessment of disease activity (PtGA) in RA and Bath Ankylosing Spondylitis Functional Index (BASFI) in AS, were revealed as the most important features in both diseases., Conclusions: RF-method exhibited superior prediction performance for responses of bDMARDs to a conventional statistical method, i.e., logistic regression, in RA patients. In contrast, despite the comparable size of the dataset, machine learning did not outperform in AS patients. The most important features of both diseases, according to feature importance analysis were patient self-reporting scales., (© 2021. The Author(s).)

We aim to generate an artificial neural network (ANN) model to predict early TNF inhibitor users in patients with ankylosing spondylitis. The baseline demographic and laboratory data of patients who visited Samsung Medical Center rheumatology clinic from Dec. 2003 to Sep. 2018 were analyzed. Patients were divided into two groups: early-TNF and non-early-TNF users. Machine learning models were formulated to predict the early-TNF users using the baseline data. Feature importance analysis was performed to delineate significant baseline characteristics. The numbers of early-TNF and non-early-TNF users were 90 and 505, respectively. The performance of the ANN model, based on the area under curve (AUC) for a receiver operating characteristic curve (ROC) of 0.783, was superior to logistic regression, support vector machine, random forest, and XGBoost models (for an ROC curve of 0.719, 0.699, 0.761, and 0.713, respectively) in predicting early-TNF users. Feature importance analysis revealed CRP and ESR as the top significant baseline characteristics for predicting early-TNF users. Our model displayed superior performance in predicting early-TNF users compared with logistic regression and other machine learning models. Machine learning can be a vital tool in predicting treatment response in various rheumatologic diseases. [ABSTRACT FROM AUTHOR]

Objectives: This study aimed to evaluate the association of knee osteoarthritis (OA) with comorbidities and health-related quality of life (HRQOL). Methods: A total of 8,907 (weighted n = 13,687,058) participants aged ≥50 years who had undergone knee radiography were selected from the 2010–2012 Korea National Health and Nutrition Examination Survey. OA was classified into four subgroups based on the presence or absence of pain and radiographic OA (ROA): non-OA (Pain-/ROA-), pain only (Pain+/ROA-), ROA only (Pain-/ROA+), and painful ROA (Pain+/ROA+). ROA was defined as Kellgren–Lawrence grade ≥ 2. HRQOL measurements including EuroQOL visual analogue scale (EQ-VAS) scores and the five dimensions and summary index of the EuroQOL-5 dimension (EQ-5D index) were also analyzed. Multivariable logistic regression and linear regression analyses were performed. Results: After adjustment for socioeconomic and lifestyle characteristics, cardiovascular disease, malignancy, and other comorbidities were not significantly associated with OA. Pain only and painful ROA were each significantly associated with limitations in physical activity (odds ratio (OR) 2.66, 95% CI 2.07–3.44, p < 0.001 and OR 2.83, 95% CI 2.25–3.58, p < 0.001, respectively), lower EQ-VAS (β-coefficient = -10.95, p < 0.001 and β-coefficient = -9.75, p < 0.001, respectively), and EQ-5D index (β-coefficient = -0.10, p < 0.001 and β-coefficient = -0.13, p < 0.001) compared with the non-OA group, whereas ROA only was not associated with limitations in physical activity or lower HRQOL score. Conclusions: Comorbidities were not significantly associated with knee OA after adjustment. Knee OA was associated with physical activity and HRQOL. Painful knee OA, with or without ROA, was more strongly associated with decreased physical activity and lower quality of life than ROA without pain. [ABSTRACT FROM AUTHOR]

Objective: To compare the clinical effectiveness of two treatment strategies for active rheumatoid arthritis (RA) refractory to conventional synthetic disease-modifying anti-rheumatic drugs (csDMARDs): starting TNF inhibitors (TNFIs) or changing csDMARDs. Methods: We used two nationwide Korean RA registries for patient selection. TNFI users were selected from the BIOPSY, which is an inception cohort of RA patients starting biologic DMARDs. As a control group, we selected RA patients with moderate or high disease activity from the KORONA database whose treatment was changed to other csDMARDs. After comparing baseline characteristics between the two groups in either unmatched or propensity score matched cohorts, we compared potential differences in the 1-year remission rate as a primary outcome and changes in HAQ-DI and EQ-5D scores as secondary outcomes. Results: A total of 356 TNFI starters and 586 csDMARD changers were identified from each registry as unmatched cohorts, and 294 patients were included in the propensity score matched cohort. In the intention-to-treat analysis, TNFI starters had higher 1-year remission rates than csDMARD changers in both unmatched (19.1 vs. 18.4%, p < 0.01) and matched cohorts (19.7 vs. 15.0%, p < 0.01). In per protocol analysis, TNFI starters had much higher remission rates in unmatched (37.2 vs. 28.0%, p = 0.04) and matched cohorts (35.4 vs. 19.1%, p = 0.04). However, in matched cohorts, no significant differences were observed between two groups in HAQ-DI and EQ-5D scores. Conclusions: We compared the clinical effectiveness of the two treatment strategies for active RA refractory to csDMARDs. TNFI starters showed higher 1-year remission rates than csDMARD changers. [ABSTRACT FROM AUTHOR]

This study aimed to evaluate the prevalence of comorbidities in patients with rheumatoid arthritis (RA) compared with the non-RA population. The 2010–2012 Korea National Health and Nutrition Examination Survey (KNHANES), which assesses the general health status of populations in South Korea using interviews and basic health assessment, was analyzed retrospectively. Weighted prevalence and odds ratio (OR) of comorbidities were analyzed in patients with RA compared with the non-RA population. The overall weighted (n = 37,453,158) prevalence of RA was 1.5%. Patients with RA were older and more female predominant than subjects without RA. The prevalence of living in an urban area, college graduation, alcohol consumption and smoking was lower in patients with RA than non-RA. Patients with RA had more comorbidities including hypertension, dyslipidemia, myocardial infarction (MI) or angina, stoke, osteoarthritis, lung cancer, colon cancer, pulmonary tuberculosis, asthma, diabetes, depression, thyroid disease and chronic kidney disease. After adjusting socioeconomic and lifestyle characteristics, RA was associated with an increased prevalence of MI or angina (OR 1.86, 95% CI 1.17–2.96, p = 0.009), pulmonary TB (OR 1.95, 95% CI 1.24–3.09, p = 0.004), asthma (OR 1.97, 95% CI 1.05–3.71, p = 0.036), thyroid disease (OR 1.71, 95% CI 1.05–2.77), depression (OR 2.38, 95% CI 1.47–3.85, p < 0.001) and hepatitis B (OR 2.34, 95% CI 1.15–4.80, p = 0.020) compared with the non-RA population. Prevalence of solid cancer was not significantly associated with RA after adjustment. [ABSTRACT FROM AUTHOR]

This study aimed to evaluate the association between rheumatoid arthritis (RA) and hearing impairment in the Korean adult population. Audiometric and laboratory test data from the 2010–2012 Korean National Health and Nutrition Examination Survey (KNHANES) were used for analysis. The relationship between RA and hearing impairment was analyzed, adjusting for various known risk factors associated with hearing impairment. RA was defined in the questionnaire as “RA diagnosed by a physician (yes/no) through a standardized interview.” We defined hearing impairment according to 2 categories of frequency (low/mid and high) as follows (average values in kHz): low/mid frequency, 0.5, 1.0, and 2.0, and high frequency, 3.0, 4.0, and 6.0. Of the subjects, 15,158 (weighted n = 32,035,996) completed the audiometric tests. The overall weighted prevalence of RA was 1.5%. The prevalence of hearing impairment was higher in the subjects with RA than in those without RA, in both, the low/mid- and high-frequency categories (21.1% vs 7.5%, p < 0.001 and 43.3% vs. 26.2%, p < 0.001, respectively). In the multivariable logistic analysis, RA (odds ratios [OR] 1.47, 95% confidence interval [CI] 1.05–2.06, p = 0.025) was an independent risk factor of low/mid-frequency hearing impairment along with age (OR 1.12, 95% CI 1.12–1.13, p < 0.001), current smoking (OR 1.27, 95% CI 1.03–1.56, p = 0.026), and college graduation (OR 0.53, 95% CI 0.39–0.72, p < 0.001). In the multivariable analysis of high-frequency hearing impairment, RA did not show any association with hearing impairment. This study suggests that RA is associated with low/mid-frequency hearing impairment after adjustment for various known risk factors. Further study is needed to verify the hearing impairment in RA. [ABSTRACT FROM AUTHOR]