Related Experiment Videos
Prediction of Post-Infectious Inflammatory Response Syndrome in Patients with Cryptococcal Meningitis Based on
Abdilahi Abdi Ibrahim1, Xiaomeng Ma1, Jia Liu1
1Department of Neurology, The Third Affiliated Hospital of Sun Yat-sen University, Guangzhou 510630, China.
Diagnostics (Basel, Switzerland)
|August 13, 2026
Summary
Predicting post-infectious inflammatory response syndrome (PIIRS) in cryptococcal meningitis patients is possible using radiomic and clinical data. Machine learning models integrating these features show high accuracy in identifying patients at risk for PIIRS.
Area of Science:
- Medical imaging analysis
- Infectious diseases
- Neurology
Background:
- Post-infectious inflammatory response syndrome (PIIRS) is a rare but serious complication of cryptococcal meningitis.
- Early identification of patients at risk for PIIRS is crucial for timely intervention and improved outcomes.
Purpose of the Study:
- To identify clinical and radiomic predictors of PIIRS in patients with cryptococcal meningitis.
- To develop and evaluate machine learning models for predicting PIIRS risk.
Main Methods:
- Retrospective analysis of 149 cryptococcal meningitis patients (65 cases with PIIRS, 84 controls).
- Extraction of 110 radiomic features from T2 FLAIR MRI scans.
- Development of machine learning models integrating radiomic, spatial, and clinical data (including C3 levels and opening pressure).
Main Results:
- Independent risk factors for PIIRS included ventriculoperitoneal shunt, elevated complement component 3 (C3), and high lumbar puncture opening pressure.
- Radiomics-based models demonstrated high predictive performance, with the best model achieving an AUC of 0.948 (test set) and 0.943 (validation set).
Conclusions:
- Radiomics-based machine learning models show promise in predicting PIIRS in cryptococcal meningitis patients.
- These findings require validation in larger, multicenter studies before clinical application.
