Establishment and Validation of a Machine-Learning Prediction Nomogram Based on Lymphocyte Subtyping for Intra-Abdominal Candidiasis in Septic Patients

Abstract

This study aimed to develop and validate a nomogram based on lymphocyte subtyping and clinical factors for the early and rapid prediction of Intra-abdominal candidiasis (IAC) in septic patients. A prospective cohort study of 633 consecutive patients diagnosed with sepsis and intra-abdominal infection (IAI) was performed. We assessed the clinical characteristics and lymphocyte subsets at the onset of IAI. A machine-learning random forest model was used to select important variables, and multivariate logistic regression was used to analyze the factors influencing IAC. A nomogram model was constructed, and the discrimination, calibration, and clinical effectiveness of the model were verified. High-dose corticosteroids receipt, the CD4⁺T/CD8⁺ T ratio, total parenteral nutrition, gastrointestinal perforation, (1,3)-β-D-glucan (BDG) positivity and broad-spectrum antibiotics receipt were independent predictors of IAC. Using the above parameters to establish a nomogram, the area under the curve (AUC) values of the nomogram in the derivation and validation cohorts were 0.822 (95% CI 0.777–0.868) and 0.808 (95% CI 0.739–0.876), respectively. The AUC in the derivation cohort was greater than the Candida score [0.822 (95% CI 0.777–0.868) vs. 0.521 (95% CI 0.478–0.563), p < 0.001]. The calibration curve showed good predictive values and observed values of the nomogram; the Decision Curve Analysis (DCA) results showed that the nomogram had high clinical value. In conclusion, we established a nomogram based on the CD4⁺/CD8⁺ T-cell ratio and clinical risk factors that can help clinical physicians quickly rule out IAC or identify patients at greater risk for IAC at the onset of infection.