Application of machine learning to identify risk factors for bloodstream infections caused by different antimicrobial resistance phenotypes in chemotherapy patients.
8 affiliations
- 1School of Arts, Sciences and Humanities, University of Sao Paulo, Sao Paulo, Brazil.
- 2Infectious Diseases Service, Cancer Institute, University of São Paulo School of Medicine, Sao Paulo, Brazil. [email protected].
- 3Department of Infectious Diseases, University of São Paulo School of Medicine, Dr Eneas de Carvalho Aguiar Street, 255, Sao Paulo, SP, 05403-900, Brazil. [email protected].
- 4Department of Hematology, Cancer Institute, University of Sao Paulo School of Medicine, Sao Paulo, Brazil.
- 5Department of Oncology, Cancer Institute, University of Sao Paulo School of Medicine, Sao Paulo, Brazil.
- 6Department of Infectious Diseases, University of São Paulo School of Medicine, Dr Eneas de Carvalho Aguiar Street, 255, Sao Paulo, SP, 05403-900, Brazil.
- 7CAMO-Net - Centre for Antimicrobial Optimisation Network, London, UK.
- 8Infectious Diseases Service, Cancer Institute, University of São Paulo School of Medicine, Sao Paulo, Brazil.
Abstract
Purpose
Bloodstream infections (BSIs) remain a major cause of morbidity and mortality in patients undergoing chemotherapy. The rising incidence of multidrug-resistant organisms (MDROs) further challenges the adequacy of empirical antimicrobial therapy. Therefore, this study aimed to identify features associated with different microbiological profiles of BSIs in chemotherapy patients using machine learning models.Methods
We retrospectively analyzed 1435 confirmed BSI episodes among 1220 patients treated at a tertiary cancer center from 2017 to 2022. A multi-label classification approach was used to model BSIs associated with carbapenem-resistant Gram-negative bacilli (CR-GNB), methicillin-resistant staphylococci, third-generation cephalosporin-resistant Enterobacterales (3GCRE), fungal pathogens, and non-MDRO/non-fungal. Classifier chain models based on logistic regression, random forests, neural networks, and gradient boosting were trained and evaluated using cross-validation with an independent test set. Model outputs were interpreted using SHapley Additive exPlanations (SHAP) to estimate feature contributions.Results
Most BSIs occurred during the first chemotherapy cycle and were due to non-MDROs or fungal agents (1193 episodes; 83.1%), CR-GNB was identified in 109 (7.6%). Model performance varied across pathogen groups, with the highest discrimination observed for CR-GNB (AUC-ROC 83.25%; 95% CI 73.91%-91.64%). Key features included hospitalization days, hematologic malignancy, prior MDRO infection, monocytopenia, and palliative chemotherapy. Chemotherapy class contributed differently across pathogen groups, with alkylating agents contributing to candidemia, antitumor antibiotics to CR-GNB, and antimetabolites to 3GCRE. Prior CR-GNB infection was the strongest predictor for CR-GNB BSI, while palliative chemotherapy correlated with non-MDRO/non-fungal BSI.Conclusion
Multi-label machine learning approaches may help disentangle the complex etiologies of BSIs in cancer patients by identifying associations between host factors, prior MDRO infection, and chemotherapy class within confirmed BSI episodes, informing future hypothesis generation.
Related
- Bloodstream Infections in Patients with Malignancy: Microbiological Epidemiology, Antimicrobial Resistance, and Factors Associated with Mortality in a Nine-Year Single-Center Retrospective Cohort
- Antimicrobial resistance in bloodstream infections in a children’s hospital, Myanmar
- Changes in the microbial spectrum and antimicrobial resistance of bloodstream infections in intensive care units before and during the COVID-19 pandemic.
- Spatiotemporal trajectories of antimicrobial resistance in bloodstream infections in China: a nationwide longitudinal surveillance study from 2014 to 2023
- Bloodstream Infections After Cardiac Surgery: Pathogen Distribution, Antimicrobial Resistance, and a Nomogram Prediction Model
BibTeX
@article{junior2026application,
title = {{Application of machine learning to identify risk factors for bloodstream infections caused by different antimicrobial resistance phenotypes in chemotherapy patients.}},
author = {Adhemar Villani Junior and Maristela P Freire and Luis Alberto De Padua Covas Lage and Felippe Lazar Neto and Maura Salaroli Oliveira and Edson Abdala and Fatima L S Nunes and Anna Sara S Levin},
journal = {Supportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer},
year = {2026},
month = oct,
doi = {10.1007/s00520-026-11379-2},
url = {https://europepmc.org/article/MED/42900238},
note = {CC BY},
}
RIS
TY - JOUR
TI - Application of machine learning to identify risk factors for bloodstream infections caused by different antimicrobial resistance phenotypes in chemotherapy patients.
AU - Adhemar Villani Junior
AU - Maristela P Freire
AU - Luis Alberto De Padua Covas Lage
AU - Felippe Lazar Neto
AU - Maura Salaroli Oliveira
AU - Edson Abdala
AU - Fatima L S Nunes
AU - Anna Sara S Levin
PY - 2026
DA - 2026/10/20
JO - Supportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer
DO - 10.1007/s00520-026-11379-2
UR - https://europepmc.org/article/MED/42900238
AB - Purpose Bloodstream infections (BSIs) remain a major cause of morbidity and mortality in patients undergoing chemotherapy. The rising incidence of multidrug-resistant organisms (MDROs) further challenges the adequacy of empirical antimicrobial therapy. Therefore, this study aimed to identify features associated with different microbiological profiles of BSIs in chemotherapy patients using machine learning models.Methods We retrospectively analyzed 1435 confirmed BSI episodes among 1220 patients treated at a tertiary cancer center from 2017 to 2022. A multi-label classification approach was used to model BSIs associated with carbapenem-resistant Gram-negative bacilli (CR-GNB), methicillin-resistant staphylococci, third-generation cephalosporin-resistant Enterobacterales (3GCRE), fungal pathogens, and non-MDRO/non-fungal. Classifier chain models based on logistic regression, random forests, neural networks, and gradient boosting were trained and evaluated using cross-validation with an independent test set. Model outputs were interpreted using SHapley Additive exPlanations (SHAP) to estimate feature contributions.Results Most BSIs occurred during the first chemotherapy cycle and were due to non-MDROs or fungal agents (1193 episodes; 83.1%), CR-GNB was identified in 109 (7.6%). Model performance varied across pathogen groups, with the highest discrimination observed for CR-GNB (AUC-ROC 83.25%; 95% CI 73.91%-91.64%). Key features included hospitalization days, hematologic malignancy, prior MDRO infection, monocytopenia, and palliative chemotherapy. Chemotherapy class contributed differently across pathogen groups, with alkylating agents contributing to candidemia, antitumor antibiotics to CR-GNB, and antimetabolites to 3GCRE. Prior CR-GNB infection was the strongest predictor for CR-GNB BSI, while palliative chemotherapy correlated with non-MDRO/non-fungal BSI.Conclusion Multi-label machine learning approaches may help disentangle the complex etiologies of BSIs in cancer patients by identifying associations between host factors, prior MDRO infection, and chemotherapy class within confirmed BSI episodes, informing future hypothesis generation.
SN - 0941-4355
SN - 1433-7339
ER -