Association rule mining identifies baseline clinical patterns associated with pulmonary infection in adults with acute leukemia: development and temporal validation of rule-derived prediction models

Objective To identify clinically interpretable baseline patterns associated with pulmonary infection in adults with acute leukemia using association rule mining and compare the predictive performance of rule-derived and conventional logistic regression models. Methods This retrospective cohort study included 1, 243 hospitalization episodes from 969 adults with acute leukemia. Episodes were chronologically divided into a development cohort (n=994) and a temporal validation cohort (n=249). Baseline demographic characteristics, treatment-related information, and laboratory measurements available within 24 h after admission were analyzed. Missing data were handled using multiple imputation within the development cohort. Association rule mining was conducted using the Apriori algorithm, and stable non-redundant rules were converted into binary indicators to develop a rule-derived logistic regression model. Its performance was compared with a conventional multivariable logistic regression model. Results Incident pulmonary infection occurred after admission in 620/994 (62.4%) episodes in the development cohort and 163/249 (65.5%) episodes in the temporal validation cohort. Five stable non-redundant association rules were identified, primarily involving induction chemotherapy, leukocytosis, neutropenia, thrombocytopenia, and elevated PCT. The strongest association was observed for induction chemotherapy combined with ANC reduction (support 0.104, confidence 0.904, lift 1.45). The rule-derived logistic regression model showed lower discrimination than the conventional logistic regression model in both the development cohort (AUC 0.760 vs 0.808) and temporal validation cohort (AUC 0.751 vs 0.781). Both models showed acceptable calibration, and decision curve analysis demonstrated positive net benefit across clinically relevant threshold probabilities. Conclusion Association rule mining identified clinically interpretable combinations of baseline characteristics associated with pulmonary infection in adults with acute leukemia. Rule-derived predictors did not improve discrimination compared with conventional logistic regression but revealed multidimensional risk patterns that may complement traditional prediction approaches. Association rule mining may therefore serve as an exploratory approach for clinical pattern discovery and hypothesis generation while complementing established prediction models.

Authors

Institutions

Publication Details

Journal
Frontiers in Oncology
Published
2026-11-17
DOI
https://doi.org/10.3389/fonc.2026.1902892
Primary Topic
Neutropenia and Cancer Infections
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Association rule mining identifies baseline clinical patterns associated with pulmonary infection in adults with acute leukemia: development and temporal validation of rule-derived prediction models

Ningting Xiao, 苗羊宝, Wen Wang, Tian Xiao et al.
2 links
Frontiers in Oncology
Neutropenia and Cancer Infections
article

Association rule mining identifies baseline clinical patterns associated with pulmonary infection in adults with acute leukemia: development and temporal validation of rule-derived prediction models

Ningting Xiao, 苗羊宝, Wen Wang, Tian Xiao, Xiaobing Huang, Bixin Huang, Jing Yang, Haiyan Wu
article en

Abstract

Objective To identify clinically interpretable baseline patterns associated with pulmonary infection in adults with acute leukemia using association rule mining and compare the predictive performance of rule-derived and conventional logistic regression models. Methods This retrospective cohort study included 1, 243 hospitalization episodes from 969 adults with acute leukemia. Episodes were chronologically divided into a development cohort (n=994) and a temporal validation cohort (n=249). Baseline demographic characteristics, treatment-related information, and laboratory measurements available within 24 h after admission were analyzed. Missing data were handled using multiple imputation within the development cohort. Association rule mining was conducted using the Apriori algorithm, and stable non-redundant rules were converted into binary indicators to develop a rule-derived logistic regression model. Its performance was compared with a conventional multivariable logistic regression model. Results Incident pulmonary infection occurred after admission in 620/994 (62.4%) episodes in the development cohort and 163/249 (65.5%) episodes in the temporal validation cohort. Five stable non-redundant association rules were identified, primarily involving induction chemotherapy, leukocytosis, neutropenia, thrombocytopenia, and elevated PCT. The strongest association was observed for induction chemotherapy combined with ANC reduction (support 0.104, confidence 0.904, lift 1.45). The rule-derived logistic regression model showed lower discrimination than the conventional logistic regression model in both the development cohort (AUC 0.760 vs 0.808) and temporal validation cohort (AUC 0.751 vs 0.781). Both models showed acceptable calibration, and decision curve analysis demonstrated positive net benefit across clinically relevant threshold probabilities. Conclusion Association rule mining identified clinically interpretable combinations of baseline characteristics associated with pulmonary infection in adults with acute leukemia. Rule-derived predictors did not improve discrimination compared with conventional logistic regression but revealed multidimensional risk patterns that may complement traditional prediction approaches. Association rule mining may therefore serve as an exploratory approach for clinical pattern discovery and hypothesis generation while complementing established prediction models.

Frontiers in OncologyVol. 16
University of Electronic Science and Technology of China (CN), Sichuan Academy of Medical Sciences & Sichuan Provincial People's Hospital (CN)
Good health and well-being
Neutropenia and Cancer Infections
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon