A Comprehensive Multimorbidity Disease Risk Prediction Framework for Irish Rheumatoid Arthritis Patients

Background Rheumatoid Arthritis (RA) is an autoimmune condition accompanied by chronic inflammation of the joints and other body tissues which affects 1% of the world population. Those who suffer from RA have a significantly higher risk of multimorbidity diseases such as Cardiovascular diseases (CVD), Major osteoporotic fractures (MOF), Covid-19 hospitalization and death. Methods The study used a dataset comprising 29,940 subjects over a period of 23 years, including 2,174 RA patients who were aged over 20 years and had undergone a DXA scan. DXA scan data were gathered from four DXA machines across three hospital sites in the West of Ireland. The study proposed the Ensemble Stacking Elastic Net (ESEN) Model for Predicting the Risk of Mortality and Multi-tasks eXtreme Gradient Boosting (MT-XGBoost) Model for Predicting the Risk of CVD, MOF, and COVID-19 hospitalization. The model predicts risk of CVD, MOF, COVID-19 hospitalization and death. Results Both models were evaluated, the ESEN model had the highest concordance index (C-Index) of 0.91 among survival analysis models. The MT-XGBoost model for binary outcomes had the highest area under the curve (AUC) for CVD (0.94), MOF (0.91), and moderate performance for COVID-19 (AUC: 0.76). Conclusion Based on data-driven methods, this research develops a first predictive model to identify RA patients who are at a higher risk of multimorbidity diseases. The findings have important implications for the clinical management of RA patients. This innovative screening tool bridges a significant gap by simultaneously predicting multiple risks, enabling the early identification of patients at heightened risk for multimorbidity outcomes.