Abstract
OBJECTIVE: This study aimed to build and validate a nomogram model to predict the risk of incomplete immune reconstitution in people living with HIV (PLWH). METHODS: Totally 3783 individuals with a confirmed diagnosis of HIV/AIDS were included. A predictive model was developed based on a retrospective set (N = 2678) and was validated using the remaining cases (N = 1105). Univariate and multivariate logistic regression analyses were performed to determine valuable predictors among the collected clinical and laboratory variables. The predictive model is presented in the form of a nomogram, which is internally and externally validated with two independent datasets. The discrimination of nomograms was assessed by calculating the area under the curve (AUC). Besides, calibration curve and decision curve (DCA) analyses were performed in the training and validation sets. RESULTS: The final model comprised 5 predictors, including baseline CD4, age at ART initiation, BMI, HZ and TBIL. The AUC of the nomogram model was 0.902, 0.926, 0.851 in the training cohort, internal validation and external cohorts. The calibration accuracy and diagnostic performance were satisfactory in both the training and validation sets. CONCLUSIONS: This predictive model based on a retrospective study was externally validated using 5 readily available clinical indicators. It showed high performance in predicting the risk of incomplete immune reconstitution in people living with HIV.