CD3+ T-cell count prediction for anti-thymocyte globulin treatment monitorization in kidney transplant recipients: a machine learning model

The study focused on developing a machine learning model to predict CD3+ T-cell counts in kidney transplant patients treated with anti-thymocyte globulin (ATG). The analysis included 397 patients, of whom 99.2% received kidneys from living donors. Results showed that 57.2% of patients achieved the target value of less than 50 cells/μl of CD3+ T-cells on day one and 57.5% of patients achieved less than 30 cells/μl on day two. The machine learning model demonstrated superior performance compared to traditional logistic regression with ROC-AUC values of 0.75 and 0.80 for day one. Researchers found that the effect of ATG treatment can be predicted using available laboratory tests and patient-specific characteristics without the need for CD3+ T-cell measurement by flow cytometry.