Key result
XGBoost model predicts unplanned readmissions in head and neck cancer with up to ~0.76 AUC.
Why the study?
Readmissions after head and neck cancer hospitalizations are common and costly, prompting the development and external validation of machine learning models to predict unplanned readmissions across short- and longer-term horizons.
Can machine learning models accurately predict 30-, 90-, and 180-day unplanned readmissions in patients hospitalized for head and neck cancer?
Population
Adult nonelective admissions with ≥1 malignant HNC diagnosis code (N = 57,201 in external test cohort)
Comparison
Machine learning prediction models based on 247 discharge time predictors
Design
Retrospective database study with external temporal validation
Follow-up
180 days
Authors
Loading...
May aid ML-guided discharge planning in head and neck cancer; leaves open prospective validation before practice change.
Observational (n=57,201)
Yes
Can machine learning models accurately predict 30-, 90-, and 180-day unplanned readmissions in patients hospitalized for head and neck cancer?
Effect estimate: AUC 0.725/0.746/0.756
Machine learning models, particularly XGBoost, can effectively predict short- and long-term unplanned readmission risk in head and neck cancer patients using administrative and clinical data.
Lee et al. (2026) conducted an observational in Head and neck cancer (n=57,201). XGBoost machine learning model vs. Treat all/none was evaluated on 30-/90-/180-day unplanned readmission (AUC 0.725/0.746/0.756). An XGBoost machine learning model effectively predicted 30-, 90-, and 180-day unplanned readmissions in patients with head and neck cancer, achieving AUCs of 0.725, 0.746, and 0.756, respectively.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: