Algorithm for calculating high disease activity in SLE.

Area of Science:

• Rheumatology
• Clinical Informatics
• Biostatistics

Background:

• Identifying patients with Systemic Lupus Erythematosus in high disease activity status (HDAS) is critical for timely treatment escalation and clinical trial selection.
• The SLEDAI (Systemic Lupus Erythematosus Disease Activity Index) is a common metric, but its absence necessitates alternative methods for assessing HDAS.
• Developing a predictive model using readily available clinical data can streamline patient stratification.

Purpose of the Study:

• To create a predictive algorithm for identifying high disease activity in lupus patients.
• To utilize simple demographic and laboratory values for model fitting, avoiding reliance on the SLEDAI score.
• To enable patient selection for treatment escalation or clinical trials based on predicted HDAS.

Main Methods:

• Prospectively collected data from 286 lupus patients over 5680 visits were analyzed.
• Combinatorial search was employed to identify algorithms predicting high disease activity (HDA), defined as SLEDAI-2K ≥10.
• Model performance was evaluated using multi-class area under the ROC curve (AUC) and compared against a naïve Bayes classifier.

Main Results:

• Sixteen laboratory parameters were significantly associated with HDA.
• A final algorithm, based on seven pathology measures and three demographic variables, achieved 88.6% accuracy and an 11.4% misclassification rate.
• The developed algorithm demonstrated superior performance (AUC = 0.829) compared to the naïve Bayes classifier (AUC = 0.663).

Conclusions:

• It is feasible to build an accurate model for calculating HDA using routinely available clinical parameters.
• The developed algorithm shows promise for identifying lupus patients with high disease activity.
• Independent validation studies are necessary to confirm the algorithm's predictive performance in diverse populations.