Dai Y, Shan W, Yang Q, et al. - In this study including individuals with tuberculosis (TB), latent TB infection (LTBI), cured TB (RxTB), pneumonia (PN) and healthy controls (HCs), researchers assessed iron homeostasis as a TB diagnostic biomarker. They established and validated their prediction model after measuring serum levels of three iron biomarkers—serum iron, ferritin, and transferrin. The correlation of the three iron biomarker levels with patient status (TB, HC, LTBI, RxTB or PN) and with the extent of lung damage and bacillary load was observed and verified in patients with TB. Then, a TB prediction model, neural network (NNET), was constructed, including the data of the three iron biomarkers. For diagnosis of TB, good performance of the model was evident. In the training data set (n = 663), the model yielded 83% sensitivity and 86% specificity. In the test data set (n = 220), the sensitivity was 70% and specificity was 92%. In the large cohort of patients examined, the differentiation of TB from HC, LTBI, RxTB and PN was allowed by the established NNET TB prediction model, which may augment current TB diagnostics.