Related Experiment Video
Updated: May 23, 2026
Murine Model of Leukemia Relapse to Induction Chemotherapy for Acute Lymphoblastic Leukemia
Published on: October 17, 2025
Machine learning-based prediction of central line-associated bloodstream infection in children with acute leukaemia
Yujie Zhang1, Yulin Wu2, Hui Gao1
1Department of Hematology, Dalian Women and Children's Medical Center (Group), Dalian, Liaoning, China.
Insights
Machine learning accurately predicts central line-associated bloodstream infections (CLABSI) in children with acute leukemia. The TabPFN model identified high-risk patients, aiding early intervention and improving outcomes.
Area of Science:
- Pediatric Oncology
- Infectious Disease Epidemiology
- Computational Biology
Background:
- Central line-associated bloodstream infection (CLABSI) is a significant complication in pediatric acute leukemia treatment.
- CLABSI adversely impacts therapeutic success and patient prognosis.
Purpose of the Study:
- To develop and evaluate a machine learning model for early prediction of CLABSI risk in pediatric acute leukemia patients.
- To identify key clinical features associated with CLABSI development.
Main Methods:
- Retrospective analysis of clinical data from 407 pediatric acute leukemia patients.
- Evaluation of six machine learning algorithms, including TabPFN, for predictive performance.
- Feature importance analysis to identify significant risk factors.
Main Results:
- The TabPFN model achieved the highest predictive accuracy at 91.2%.
- Key predictors for CLABSI included corticosteroid type, body temperature, neutrophil count, and white blood cell count.
- The model demonstrated potential for effective CLABSI risk stratification.
Conclusions:
- Machine learning models, particularly TabPFN, show promise as decision-support tools for managing CLABSI risk in pediatric leukemia.
- Further validation is necessary for widespread clinical implementation of these predictive models.
Background:
Central line-associated bloodstream infection (CLABSI) is a common and serious complication in children with acute leukaemia (AL), leading to increased morbidity, prolonged hospitalization, and adverse clinical outcomes. Early identification of patients at high risk of CLABSI remains a major clinical challenge.
Aim:
To develop and evaluate machine learning models for early prediction of CLABSI risk in paediatric patients with AL and to identify key clinical factors associated with infection.
Methods:
A retrospective study was conducted using clinical data from 407 paediatric patients with AL. Clinical variables were collected and preprocessed for model development. Six machine learning algorithms were constructed and compared for CLABSI prediction. Model performance was evaluated using standard classification metrics, and feature importance analysis was performed to identify major predictors associated with CLABSI.
Findings:
Among the evaluated models, the Tabular Prior-Fitted Network (TabPFN) achieved the best predictive performance, with an accuracy of 91.2%. Feature importance analysis indicated that corticosteroid type, body temperature, neutrophil count, and white blood cell count were the most influential factors associated with CLABSI risk. The results demonstrated that machine learning models can effectively distinguish patients at high risk of infection.
Conclusion:
Machine learning-based prediction models show considerable potential for early CLABSI risk assessment in paediatric patients with AL. The proposed approach may support clinical decision-making and facilitate timely preventive interventions. Further multicentre studies are required to validate the model and assess its generalizability in broader clinical settings.
