Development and validation of a multivariate predictive model for rheumatoid arthritis mortality using a machine...

Area of Science:

• Rheumatology
• Biostatistics
• Machine Learning

Background:

• Rheumatoid arthritis (RA) is a chronic autoimmune disease associated with increased mortality.
• Accurate prediction of RA mortality is crucial for patient management and resource allocation.

Purpose of the Study:

• To develop and validate a machine learning-based clinical prediction model for rheumatoid arthritis mortality.
• To identify key demographic and clinical predictors of RA mortality.

Main Methods:

• Utilized Random Survival Forests (RSF), a machine learning technique, for model development.
• Trained the model on the Hospital Clínico San Carlos RA Cohort (HCSC-RAC) and validated it on the Hospital Universitario de La Princesa Early Arthritis Register Longitudinal study (PEARL).
• Included demographic and clinical variables from the first two years post-diagnosis.

Main Results:

• Age at diagnosis, median erythrocyte sedimentation rate, and number of hospital admissions were identified as significant predictors.
• The model achieved prediction errors of 0.187 (training) and 0.233 (validation).
• Five distinct mortality risk groups were identified, with time-dependent sensitivity and specificity ranging from 0.43-0.80 in the validation cohort.

Conclusions:

• A robust clinical prediction model for RA mortality was successfully developed and validated using RSF.
• The model demonstrates potential for identifying patients at higher risk of mortality.
• Further external validation is recommended to confirm the model's generalizability.