Related Experiment Video
Updated: Jan 17, 2026
Detection of Invasive Pulmonary Aspergillosis in Haematological Malignancy Patients by using Lateral-flow Technology
Published on: March 22, 2012
Development of a Machine-Learning Immuno-Serologic Diagnostic Model for Non-Neutropenic Invasive Pulmonary Fungal
Hui Huang1,2, Fang Fang3, Weiguo Lu2
1The First Clinical Medical College, Guangzhou University of Chinese Medicine, Guangzhou, Guangdong, People's Republic of China.
Background:
Non-neutropenic invasive pulmonary fungal disease (IPFD) is increasingly recognized but remains challenging to diagnose due to nonspecific clinical and radiological features. This retrospective, single-center study was conducted at the First Affiliated Hospital of Guangzhou University of Chinese Medicine and aimed to develop and evaluate a diagnostic model based on immuno-Serologic biomarkers for distinguishing non-neutropenic IPFD from bacterial pneumonia.
Methods:
A total of 157 pneumonia patients (65 non-neutropenic IPFD cases and 92 bacterial pneumonia cases) admitted to the First Affiliated Hospital of Guangzhou University of Chinese Medicine between April 2018 and December 2022 were enrolled. Least Absolute Shrinkage and Selection Operator (LASSO) regression and collinearity analysis were applied to screen key variables, followed by the development of diagnostic models using nine machine learning algorithms. Model performance was comprehensively evaluated, and temporal validation in an independent later cohort from the same center was conducted using data from 102 pneumonia patients (33 non-neutropenic IPFD and 69 bacterial pneumonia cases) admitted between January 2023 and March 2025.
Results:
Five biomarkers were identified as predictors: galactomannan (GM), monocyte human leukocyte antigen-DR expression (mHLA-DR), monocyte count, interleukin-6 (IL-6), and 1,3-β-D-glucan (BDG). The Light Gradient Boosting Machine (LightGBM) model demonstrated optimal performance in the validation set, with an area under the receiver operating characteristic curve (AUC) of 0.865 (95% CI: 0.728-0.999) and accuracy of 0.781. In the test set, the model achieved an AUC of 0.810 and accuracy of 0.750. Decision curve analysis (DCA) indicated favorable net benefits across probability thresholds of 0-1. Temporal validation yielded an AUC of 0.821 and accuracy of 0.794.
Conclusion:
The immuno-serologic diagnostic model exhibits strong discriminatory performance in differentiating bacterial pneumonia from non-neutropenic IPFD, highlighting its potential for clinical application.
