EUVIMEDEUVIMED entdecken
ValidierungsstudieOpen Access · CC BY-NC 4.0

Development and validation of an explainable machine learning model for predicting prognosis in sepsis patients with a history of cancer who were admitted to the intensive care unit

Entwicklung und Validierung eines erklärbaren Machine-Learning-Modells zur Prognose bei Sepsispatient:innen mit Krebsanamnese auf der Intensivstation

Xiang Luo, Xiuji Kan iD, Dongliang Wang, Yu Shi, Siqi Zhu, Zhenyu Chen, Congcong Wang, Wenqi Zhu, Xiangtong Wang, Wenwen Sun

Journal of International Medical Research · 2025-08 · 53 (8)

DOI: 10.1177/03000605251362991 ↗

Auf einen Blick

Eine retrospektive Studie entwickelte ein Machine-Learning-Modell zur Vorhersage der Krankenhaussterblichkeit bei Sepsispatient:innen mit Krebsanamnese. Das Random-Forest-Modell zeigte die beste Vorhersageleistung.

Fragestellung

Welche Machine-Learning-Modelle sagen die In-Hospital-Mortalität bei Sepsispatient:innen mit Krebsanamnese auf der Intensivstation am besten voraus?

Methode

Retrospektive Analyse von EHR-Daten aus MIMIC-IV; 8 Machine-Learning-Algorithmen, 5-fach-Validierung; Interpretierbarkeit mit SHAP, Partial Dependence Plots und break down.

Ergebnisse

3364 Patient:innen. Random Forest: AUC 0,78 (95 %-KI 0,76–0,80), Brier-Score 0,15; als bestes Modell unter 8 getesteten Algorithmen.

Schlussfolgerung der Autor:innen

Die Autor:innen entwickelten Machine-Learning-Modelle zur Vorhersage der In-Hospital-Mortalität bei Sepsispatient:innen mit Krebsanamnese und nutzten MIMIC-IV für eine umfassende Risikobewertung.

Kernaussagen

  • 3364 Patient:innen mit Sepsis und Krebsanamnese auf der Intensivstation
  • Random Forest mit höchster AUC 0,78 (95 %-KI 0,76–0,80)
  • Explanable-ML mit SHAP, Partial Dependence Plots und break down

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

Kurz gefasst

Validierungsstudie in der Zeitschrift Journal of International Medical Research, online erschienen am 13.08.2025. Verfasst von Xiang Luo und 9 weiteren Autor:innen. Beteiligte Einrichtung: Jiangsu Province (Suqian) Hospital, Nanjing Medical University, China. Hauptthemen laut MEDLINE: Machine Learning, Sepsis, Intensive Care Units, Neoplasms. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: Jiangsu Province (Suqian) Hospital 136 Talent training plan, Jiangsu Province (Suqian) Hospital Research Project.

Auf einen BlickAbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Sepsis is the leading cause of mortality in critically ill cancer patients; however, traditional prognostic models fail to capture the complexity of their immune and physiological interactions.

Methods

This retrospective study analyzed electronic health records from the Medical Information Mart for Intensive Care IV database, including the records of patients with sepsis who had a documented history of cancer and were admitted to the intensive care unit. A two-step feature selection approach, combining least absolute shrinkage and selection operator regression and recursive feature elimination, was used to identify key prognostic variables. Eight machine learning algorithms, such as random forest and extreme gradient boosting, were trained and evaluated using five-fold cross-validation. Model performance was assessed using the area under the receiver operating characteristic curve value, Brier scores, sensitivity, and specificity. SHapley Additive exPlanations, Partial Dependence Plots, and break down algorithms were applied to enhance model interpretability.

Results

The final cohort included 3364 patients admitted to the intensive care unit. Nonsurvivors had significantly higher illness severity scores (Acute Physiology Score III and Sequential Organ Failure Assessment) than survivors (p < 0.001). Among the tested models, the random forest model demonstrated superior performance, achieving the highest area under the receiver operating characteristic curve value (0.78; 95% confidence interval: 0.76–0.80) and the lowest Brier score (0.15), indicating strong predictive accuracy.

Conclusions

This study developed machine learning models for predicting in-hospital mortality in sepsis patients with a history of cancer, leveraging the Medical Information Mart for Intensive Care IV database for comprehensive risk assessment.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Machine LearningSepsisIntensive Care UnitsNeoplasmsHumansPrognosisMaleFemaleRetrospective StudiesMiddle AgedAgedROC CurveAlgorithmsElectronic Health RecordsHospital MortalitySeverity of Illness IndexOrgan Dysfunction Scores

Stichwörter der Autor:innen

Machine learningcancerprediction modelsepsissupervised learning

Autor:innen und Institutionen

  • Jiangsu Province (Suqian) Hospital, Nanjing Medical University, China.
  • First Affiliated Hospital of Soochow University, Soochow University, China.
  • General Hospital of Eastern Theater Command of the People's Liberation Army, China.

ORCID-Profile: Xiuji Kan

Förderung

  • Jiangsu Province (Suqian) Hospital 136 Talent training plan
  • Jiangsu Province (Suqian) Hospital Research Project

Zitieren

APA

Luo, X., Kan, X., Wang, D., Shi, Y., Zhu, S., Chen, Z., Wang, C., Zhu, W., Wang, X., & Sun, W. (2025). Development and validation of an explainable machine learning model for predicting prognosis in sepsis patients with a history of cancer who were admitted to the intensive care unit. Journal of International Medical Research, 53(8). https://doi.org/10.1177/03000605251362991

Vancouver

Luo X, Kan X, Wang D, Shi Y, Zhu S, Chen Z, et al. Development and validation of an explainable machine learning model for predicting prognosis in sepsis patients with a history of cancer who were admitted to the intensive care unit. Journal of International Medical Research. 2025;53(8). doi:10.1177/03000605251362991

BibTeX
@article{Luo2025,
  author = {Luo, Xiang and Kan, Xiuji and Wang, Dongliang and Shi, Yu and Zhu, Siqi and Chen, Zhenyu and Wang, Congcong and Zhu, Wenqi and Wang, Xiangtong and Sun, Wenwen},
  title = {Development and validation of an explainable machine learning model for predicting prognosis in sepsis patients with a history of cancer who were admitted to the intensive care unit},
  journal = {Journal of International Medical Research},
  year = {2025},
  volume = {53},
  number = {8},
  doi = {10.1177/03000605251362991}
}
Kostenlos starten →