Predictive Model Shows Moderate Performance for Psoriasis Relapse Risk

TUESDAY, April 28, 2026 (HealthDay News) -- A predictive model for psoriasis relapse risk demonstrates moderate performance, according to a study published online April 11 in Scientific Reports.Xiaoxue Zhang, from The Second Affiliated Hospital of Hunan University of Traditional Chinese Medicine in Changsha, China, and colleagues developed and validated a risk prediction model for psoriasis relapse in a sample of 504 patients with psoriasis admitted to a tertiary hospital between January 2022 and December 2024. A total of 353 cases were included in the training set and 151 were included in the testing set. A univariate analysis was used to identify independent risk factors for psoriasis relapse, and a prediction model was developed using logistic regression analysis.The researchers found that the one-year relapse rate of psoriasis was 66.67 percent after treatment. Six independent risk factors were identified for psoriasis relapse: body mass index, diabetes, biologic use, smoking, upper respiratory tract infection, and nonstandard medication; these factors were all incorporated into the model. For the training and testing sets, the area under the receiver operating characteristic curve values were 0.767 and 0.704, respectively. Moderate discrimination and good calibration were seen with the model. Clinically meaningful net benefit was confirmed by a decision curve analysis in both the training and test sets."Rigorous external validation in independent, multicenter large-cohort studies is necessary before applying this model in clinical practice," the authors write.Abstract/Full Text.Sign up for our weekly HealthDay newsletter

Predictive Model Shows Moderate Performance for Psoriasis Relapse Risk

Related Stories

Swept-Source Optical Coherence Tomography Angiography May Predict Diabetic Nephropathy

Machine Learning Predicts Asthma Risk in Children With Early-Life Atopic Dermatitis

Chatbots Do Not Perform Well for Misinformation-Prone Health Topics

Large Language Models Perform Poorly for Differential Diagnosis