AI-delirium guard: Predictive modeling of postoperative delirium in elderly surgical patients
AI-delirium guard: Prädiktive Modellierung des postoperativen Delirs bei älteren chirurgischen Patient:innen
PLOS One · 05.06.2025 · 20 (6) · e0322032
DOI: 10.1371/journal.pone.0322032 ↗Auf einen Blick
Bei 17.000 Patient:innen über 65 wurde ein XGBoost-Modell zur Vorhersage eines postoperativen Delirs entwickelt. Das Modell erreichte eine AUC von 0,85.
Fragestellung
Kann ein Machine-Learning-Modell ein postoperatives Delir bei älteren chirurgischen Patient:innen vorhersagen?
Methode
Analyse von NSQIP-Daten mit 17.000 Patient:innen über 65; Vergleich mehrerer Machine-Learning-Verfahren; Auswahl des Modells nach ROC-AUC; Hyperparameter-Tuning für XGBoost.
Ergebnisse
Optimiertes XGBoost-Modell: AUC 0,85, Accuracy 0,926, Recall 0,945, Precision 0,934, F1-Score 0,939.
Einschränkungen
Weitere Untersuchungen sollen die Integration in klinische Abläufe und eine externe Validierung prüfen.
Schlussfolgerung der Autor:innen
Die Autor:innen schlagen ein starkes XGBoost-Modell zur Vorhersage von POD bei älteren chirurgischen Patient:innen vor.
Kernaussagen
- XGBoost erreichte eine AUC von 0,85
- Datenbasis: 17.000 Patient:innen über 65 aus NSQIP
- Externe Validierung und klinische Integration werden gefordert
Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.
Kurz gefasst
Fachartikel in der Zeitschrift PLOS One, online erschienen am 05.06.2025. Verfasst von Sri Harsha Boppana und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Internal Medicine, Nassau University Medical Center, East Meadow, New York, United States of America. Hauptthemen laut MEDLINE: Machine Learning, Delirium, Postoperative Complications. Die Arbeit wurde bisher 11-mal zitiert (Crossref) und nennt 49 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Themen: Delir
Abstract (Originalsprache)
Introduction
In older patients, postoperative delirium (POD) is a major complication that can result in greater morbidity, longer hospital stays, and higher healthcare expenses. Accurate prediction models for POD can enhance patient outcomes by guiding preventative strategies. This study utilizes advanced machine learning techniques to develop a predictive model for POD using comprehensive perioperative data.
Methods
We examined information from the National Surgical Quality Improvement Program (NSQIP), which included 17,000 patients who were over 65 and undergoing different types of surgery. The dataset included variables such as patient demographics (age, sex), comorbidities (diabetes, cardiovascular diseases, pre-existing dementia), surgical details (type, duration), anesthesia type and dosage, and postoperative outcomes. Categorical variables were encoded numerically, and data standardization was applied to ensure normal distribution. A range of machine learning approaches were assessed such as Decision Trees and Random Forests. Based on the greatest Area Under the Curve (AUC) from Receiver Operating Characteristic (ROC) analysis, the final model was chosen. Hyperparameter tuning was performed using GridSearchCV, optimizing parameters like max_depth, min_child_weight, and gamma for XGBoost model.
Results
The optimized XGBoost model demonstrated superior performance, achieving an AUC of 0.85. Key hyperparameters included min_child_weight = 1, max_depth = 5, gamma = 0.3, subsample = 0.9, colsample_bytree = 0.7, reg_alpha = 0.0007, learning_rate = 0.14, and n_estimators = 123. The model exhibited an accuracy of 0.926, recall of 0.945, precision of 0.934, and an F1-score of 0.939, depicting a higher level of predictive accuracy & balance between sensitivity and specificity.
Conclusion
This study proposes a strong XGBoost-based model to predict POD in older surgical patients, demonstrating the potential of Machine Learning (ML) in clinical risk assessment. With the help of the model’s balanced performance indicators and high accuracy, physicians may identify high-risk patients and promptly execute interventions in clinical settings. Subsequent investigations ought to concentrate on integration into clinical workflows and external validation.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- Department of Internal Medicine, Nassau University Medical Center, East Meadow, New York, United States of America.
- Department of Applied Physics, Delhi Technological University, Delhi, India.
- Department of Computer Science, Florida International University, Miami, Florida, United States of America.
- Department of Internal Medicine, Alluri Sitarama Raju Academy of Medical Sciences, Eluru, India.
- Johns Hopkins University Zanvyl Krieger School of Arts and Sciences, Baltimore, Maryland, United States of America.
- Department of Anesthesiology & Critical Care Medicine, Johns Hopkins School of Medicine, Baltimore, Maryland, United States of America.
ORCID-Profile: Sachin Komati, Ritwik Raj, C David Mintz
Zitieren
APA
Boppana, S. H., Tyagi, D., Komati, S., Boppana, S. L., Raj, R., & Mintz, C. D. (2025). AI-delirium guard: Predictive modeling of postoperative delirium in elderly surgical patients. PLOS One, 20(6), e0322032. https://doi.org/10.1371/journal.pone.0322032
Vancouver
Boppana SH, Tyagi D, Komati S, Boppana SL, Raj R, Mintz CD. AI-delirium guard: Predictive modeling of postoperative delirium in elderly surgical patients. PLOS One. 2025;20(6):e0322032. doi:10.1371/journal.pone.0322032
BibTeX
@article{Boppana2025,
author = {Boppana, Sri Harsha and Tyagi, Divyansh and Komati, Sachin and Boppana, Sri Lasya and Raj, Ritwik and Mintz, C David},
title = {AI-delirium guard: Predictive modeling of postoperative delirium in elderly surgical patients},
journal = {PLOS One},
year = {2025},
volume = {20},
number = {6},
pages = {e0322032},
doi = {10.1371/journal.pone.0322032}
}
EUVIMED