Using Machine Learning Model to Predict Anxiety in Systemic Lupus Erythematosus Patients

Introduction: Systemic Lupus Erythematosus (SLE) is a chronic autoimmune disease that affects multiple organs and significantly impacts quality of life, particularly in women, with a global incidence of 5.14 cases per 100,000 person-years. Many SLE patients experience psychiatric complications, such as anxiety, with 55.4% affected in Morocco. To improve patient management and early detection of anxiety, a study proposes developing a machine learning algorithm to analyze patient data and identify predictive patterns. Method: A cross-sectional study conducted at the Hassan II University Hospital examined adults with systemic lupus erythematosus (SLE) between 2020 and 2023, excluding those with psychiatric disorders or communication problems. Data was collected through questionnaires assessing socio-demographic and clinical factors, including anxiety levels, using the Moroccan HADS. Machine learning techniques were used to analyse the data and predict anxiety. Results: The study found a significant positive association between anxiety and factors such as age, age of diagnosis, symptom duration, marital status, pre-existing conditions and respiratory problems. Model performance varied, with logistic regression achieving the highest accuracy (0.67) and recall (0.90), while random forest had the lowest accuracy (0.57) and recall (0.50). Precision scores ranged from 0.54 for SVM to 0.60 for logistic regression and decision tree, with F1 scores between 0.53 and 0.72. Conclusion: The study highlights the need to predict and treat anxiety in lupus patients, using logistic regression as an effective. Addressing anxiety through targeted care strategies can significantly improve the quality of life and overall well-being of lupus patients.