Personalized prediction of psoriasis relapse post-biologic discontinuation: a machine learning-driven population...

Area of Science:

• Dermatology
• Clinical Research
• Predictive Modeling

Background:

• Psoriasis relapse after biologic discontinuation poses a significant challenge.
• Optimizing treatment strategies can reduce relapse rates and disease burden.

Purpose of the Study:

• To develop and validate a personalized prediction model for psoriasis relapse.
• Aid in optimizing treatment strategies for patients discontinuing biologics.

Main Methods:

• Utilized Boruta algorithm and multivariate Cox regression to identify relapse predictors.
• Developed a nomogram and online calculator for outcome visualization and computation.
• Assessed model performance using ROC curves, AUC, C-statistics, calibration plots, and DCA.

Main Results:

• Included 597 patients (534 derivation, 63 validation).
• Identified anxiety, disease duration, prior/treatment duration, time to PASI 75, and max PASI as key predictors.
• Model demonstrated good predictive accuracy in internal and external validation.

Conclusions:

• A multivariate model using clinical data can predict psoriasis relapse risk post-biologic discontinuation.
• The model supports personalized treatment strategies for managing psoriasis.