2026-10-13T03:17:13Z https://www.scidb.cn/oai

10.57760/sciencedb.j00253.05010 2026-05-29T16:45:49Z

2026-05-29 Application of Machine Learning Models Based on Systemic Immune Profiles in the Differential Diagnosis of Autoimmune Diseases doi:10.57760/sciencedb.j00253.05010 en Objective To explore the value of machine learning models based on peripheral blood immune profiles in the differential diagnosis of four autoimmune diseases (SLE, RA, pSS, AS) and non-autoimmune joint diseases.Methods The multidimensional clinical data of 639 patients (525 with AIDs and 114 controls) were retrospectively analyzed. Six machine learning classifiers were constructed, and the SHAP algorithm was utilized to interpret the optimal model.Results Significant differences in immune characteristics were observed among the disease groups (P<0.05). The XGBoost model demonstrated the highest classification performance (AUC=0.998, accuracy=0.982). SHAP analysis extracted five core classification biomarkers, including Treg/CD4+(%), Treg(%), and anti-cardiolipin antibody IgG.Conclusions The XGBoost multi-class classification model based on systemic immune profiles exhibits high accuracy, holding promise as an objective tool for early auxiliary diagnosis in rheumatology. Autoimmune diseases; Differential diagnosis; Systemic immune profiles; Machine learning; XGBoost; SHAP algorithm Zhaoda.Yu Ming.Zong Shibo.Cao Zicheng.Yang Jixiao.Zhu PUBLIC https://creativecommons.org/publicdomain/zero/1.0/ dataset Science Data Bank