Chakri Imad, Qarmiche Noura, Omari Mohammed, Tachfouti Nabil, El Fakir Samira, Otmani Nada
Medical Informatics, Laboratory of Epidemiology, Biostatistics and Health Information Processing, Agadir, MAR.
Biostatistics-Informatics Unit, Department of Epidemiology, Clinical Research and Community Health, Faculty of Medicine, Pharmacy and Dentistry, Sidi Mohamed Ben Abdellah University, Fez, MAR.
Cureus. 2025 Mar 17;17(3):e80723. doi: 10.7759/cureus.80723. eCollection 2025 Mar.
Background and objective Rheumatoid arthritis (RA) is a chronic inflammatory condition that significantly impacts the quality of life. Depression in RA exacerbates pain and reduces the likelihood of remission. Predicting depression in RA is often neglected due to time and resource constraints. Hence, this study aimed to develop a machine learning (ML) model for predicting depression in RA patients. Methodology We included 112 RA patients from CHU Hassan II, Fez, Morocco. Depression was assessed using the Hospital Anxiety and Depression Scale (HADS) scale, and clinical data were extracted from medical records. Twelve features were used to develop five ML models: support vector machine (SVM), random forest (RF), decision tree (DT), logistic regression (LR), and gradient boosting classifier (GBC). Data preprocessing involved managing missing values, normalizing data, and encoding variables. Model performance was evaluated using accuracy, precision, recall, F1 score, and area under the curve (AUC). Results The ML-based feature selection method showed the optimal performance. The LR model performed best in predicting depression, with 76.5% accuracy, 72.2% precision, 81.2% recall, an F1 score of 0.765, and an area under the receiver operating characteristic curve (ROC AUC) of 0.767. Conclusions Our study highlights the significance of ML models in predicting depression in RA patients. The selected features and the LR model showed promising performance. Further research is required to validate these results and develop more advanced models. Utilizing such tools could significantly impact the management of RA patients by identifying those at risk of depression and providing appropriate psychological support.
背景与目的 类风湿关节炎（RA）是一种慢性炎症性疾病，对生活质量有显著影响。类风湿关节炎患者的抑郁会加重疼痛并降低缓解的可能性。由于时间和资源限制，类风湿关节炎患者抑郁的预测常常被忽视。因此，本研究旨在开发一种用于预测类风湿关节炎患者抑郁的机器学习（ML）模型。方法 我们纳入了来自摩洛哥非斯市哈桑二世大学医院中心（CHU Hassan II）的112例类风湿关节炎患者。使用医院焦虑抑郁量表（HADS）评估抑郁情况，并从医疗记录中提取临床数据。使用12个特征开发5种机器学习模型：支持向量机（SVM）、随机森林（RF）、决策树（DT）、逻辑回归（LR）和梯度提升分类器（GBC）。数据预处理包括处理缺失值、归一化数据和对变量进行编码。使用准确率、精确率、召回率、F1分数和曲线下面积（AUC）评估模型性能。结果 基于机器学习的特征选择方法表现出最佳性能。逻辑回归模型在预测抑郁方面表现最佳，准确率为76.5%，精确率为72.2%，召回率为81.2%，F1分数为0.765，受试者工作特征曲线下面积（ROC AUC）为0.767。结论 我们的研究强调了机器学习模型在预测类风湿关节炎患者抑郁方面的重要性。所选特征和逻辑回归模型表现出良好的性能。需要进一步研究来验证这些结果并开发更先进的模型。利用此类工具可通过识别有抑郁风险的患者并提供适当的心理支持，对类风湿关节炎患者的管理产生重大影响。
