Machine Learning Model Accurately Predicts Hyperglycemia Risk in Patients With Psoriasis
A machine learning–based predictive model may help dermatologists identify psoriasis patients at higher risk for hyperglycemia, supporting more personalized treatment planning and metabolic risk monitoring. In a validation study comparing multiple algorithms, an extreme gradient boosting (XGBoost) model demonstrated the strongest and most consistent performance across internal and external datasets.
Investigators analyzed clinical data from 575 patients with psoriasis admitted to a dermatology department, randomly dividing the cohort into training and internal test sets. External validation was performed using data from 135 patients with psoriasis identified in the National Health and Nutrition Examination Survey. Eleven machine learning approaches were evaluated, including tree-based models, neural networks, support vector machines, and traditional regression techniques.
XGBoost emerged as the top-performing model, with area under the receiver operating characteristic curve values of 0.821 in the training set, 0.820 in the internal test set, and 0.788 in the external test set. The authors reported that calibration curves and clinical decision curve analyses supported both accuracy and potential clinical usefulness across populations.
“The XGBoost-based model effectively predicts hyperglycemia risk in psoriasis patients,” the authors wrote, highlighting its ability to maintain performance in an external cohort. To facilitate clinical application, the team developed a web-based calculator, allowing clinicians to estimate individual patient risk using readily available clinical inputs.
The study addresses a growing clinical concern. Psoriasis is increasingly recognized as a systemic inflammatory disease with metabolic comorbidities, including insulin resistance and hyperglycemia. Early identification of patients at elevated metabolic risk may allow for closer monitoring, earlier intervention, or consideration of treatment strategies that minimize metabolic burden.
The authors concluded that the model “emphasiz[es] personalized treatment plans for high-risk individuals to manage hyperglycemia progression and psoriasis-related inflammation.” While the tool is not intended to replace clinical judgment, it may support decision-making in complex patients with overlapping inflammatory and metabolic risk profiles.
Reference
Hu M, Chen D, Yu J. Predictive modeling of the risk of hyperglycemia in psoriasis patients using machine learning: a multicenter retrospective study. Clin Cosmet Investig Dermatol. 2025;18:3667-3680. doi:10.2147/CCID.S552796
