Development and external evaluation of an interpretable machine-learning model for early prediction of organ failure in higher-risk acute pancreatitis patients: A multicentre cohort study
Entwicklung und externe Evaluation eines interpretierbaren Machine-Learning-Modells zur Früherkennung von Organversagen bei Patient:innen mit höherem Risiko bei akuter Pankreatitis: multizentrische Kohortenstudie
PLOS Digital Health · 22.09.2026 · 5 (9) · e0001735
DOI: 10.1371/journal.pdig.0001735 ↗Auf einen Blick
In zwei Kohorten mit akuter Pankreatitis sagte XGBoost ein neu aufgetretenes Organversagen innerhalb von 28 Tagen am besten voraus. Die externe Diskriminierung war gut.
Fragestellung
Lassen sich neue Organversagen bei akuter Pankreatitis bei Aufnahme mit Machine-Learning-Modellen vorhersagen?
Methode
Multizentrische Kohortenstudie mit Entwicklung in MIMIC-IV und externer Evaluation in zwei chinesischen Kliniken; Modellierung mit fünf Machine-Learning-Modellen und einem MLP-Benchmark, ohne Update.
Ergebnisse
In der Entwicklungskoorte entwickelten 341/1429 Patient:innen (23,9 %) ein Organversagen, in der Validierungskohorte 45/216 (20,8 %). XGBoost erzielte extern eine AUC von 0,837 (95 %-KI 0,771–0,902).
Einschränkungen
Modellauswahl, Schwellenwertwahl und Rekalibrierung nutzten die Evaluationskohorte; das Modell braucht eine unabhängige Kohorte. Prospektive Usability- und Impact-Studien fehlen.
Schlussfolgerung der Autor:innen
XGBoost sagte neues Organversagen mit guter Diskriminierung in einer schwereren AP-Kohorte voraus; SHAP machte Einzelfallvorhersagen interpretierbar.
Kernaussagen
- XGBoost zeigte extern die beste Diskriminierung
- AUC 0,837 (95 %-KI 0,771–0,902)
- SHAP visualisierte Entscheidungen und Einzelfallvorhersagen
Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.
Kurz gefasst
Kohortenstudie in der Zeitschrift PLOS Digital Health, online erschienen am 22.09.2026. Verfasst von Di Wu und 17 weiteren Autor:innen. Beteiligte Einrichtung: Pancreas Center, First Affiliated Hospital with Nanjing Medical University, Nanjing, Jiangsu, China. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 37 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: China Postdoctoral Science Foundation.
Abstract (Originalsprache)
Organ failure (OF) is the most important determinant of prognosis in acute pancreatitis (AP) and its duration defines disease severity. Early identification of individuals at high risk of developing OF is crucial. We aimed to develop machine-learning models, benchmarked against a feedforward multilayer-perceptron (MLP) neural network, to predict new-onset OF at admission using static admission-day variables. In this study, data were extracted from MIMIC-IV (development cohort) and a multicentre AP cohort from two Chinese tertiary teaching hospitals (evaluation cohort), which excluded mild AP and therefore comprised patients requiring ICU-level or high-dependency care. Features were selected with Boruta and LASSO. Five machine-learning models and one multilayer-perceptron benchmark were developed in MIMIC-IV and externally evaluated in the Xiangya cohort without updating. SHapley Additive exPlanations (SHAP) were used to visualize decision-making patterns and individual prediction interpretations. The best-performing model was deployed as an interactive, web-based tool. We found that in the discovery cohort, 341 (23.9%) of 1429 AP patients developed new-onset OF within 28 days of admission while 45 of 216 patients (20.8%) in the validation cohort developed OF. Boruta and LASSO algorithms identified six key predictors including blood urea nitrogen, platelets, triglyceride-glucose index, albumin, white blood cells, and partial thromboplastin time, which were used to construct the ML and MLP models. XGBoost gave the best discrimination on external evaluation (AUC 0.837, 95% CI 0.771–0.902). Logistic recalibration improved calibration, and the recalibrated XGBoost model was implemented as a web-based research prototype (Decent app). In conclusion, XGBoost predicted new-onset OF with good discrimination in a severity-enriched AP cohort, and SHAP made individual predictions interpretable. Model selection, threshold selection and recalibration all used the evaluation cohort, so the deployed model still requires an independent cohort. Prospective usability and clinical-impact studies are needed before clinical use.
Autor:innen und Institutionen
- Pancreas Center, First Affiliated Hospital with Nanjing Medical University, Nanjing, Jiangsu, China.
- Department of Geriatric Gastroenterology, Xiangya Hospital, Central South University, Changsha, Hunan, China.
- Institute of Systems, Molecular and Integrative Biology, University of Liverpool, Liverpool, United Kingdom.
- Emergency Department, the Third Xiangya Hospital, Central South University, Changsha, Hunan, China.
- West China Centre of Excellence for Pancreatitis, Institute of Integrated Traditional Chinese and Western Medicine, West China-Liverpool Biomedical Research Centre, Chengdu, Sichuan, China.
- Pediatric Hematology Laboratory, Division of Hematology/Oncology, Department of Pediatrics, The Seventh Affiliated Hospital of Sun Yat-Sen University, Shenzhen, Guangdong, China.
ORCID-Profile: Di Wu, Anthony Evans
Förderung
- China Postdoctoral Science Foundation
Zitieren
APA
Wu, D., Cai, W., Chen, C., Chen, M., Lv, Y., Huang, Y., Evans, A., Lin, J., Latawiec, D., Kattakayam, A., Mukherjee, R., Huang, W., Xia, Q., Xiao, J., Su, C., Peng, J., Jiang, K., & Sutton, R. (2026). Development and external evaluation of an interpretable machine-learning model for early prediction of organ failure in higher-risk acute pancreatitis patients: A multicentre cohort study. PLOS Digital Health, 5(9), e0001735. https://doi.org/10.1371/journal.pdig.0001735
Vancouver
Wu D, Cai W, Chen C, Chen M, Lv Y, Huang Y, et al. Development and external evaluation of an interpretable machine-learning model for early prediction of organ failure in higher-risk acute pancreatitis patients: A multicentre cohort study. PLOS Digital Health. 2026;5(9):e0001735. doi:10.1371/journal.pdig.0001735
BibTeX
@article{Wu2026,
author = {Wu, Di and Cai, Wenhao and Chen, Chunmei and Chen, Minting and Lv, Yang and Huang, Yilin and Evans, Anthony and Lin, Juan and Latawiec, Diane and Kattakayam, Arjun and Mukherjee, Rajarshi and Huang, Wei and Xia, Qing and Xiao, Jie and Su, Chunqiu and Peng, Jie and Jiang, Kuirong and Sutton, Robert},
title = {Development and external evaluation of an interpretable machine-learning model for early prediction of organ failure in higher-risk acute pancreatitis patients: A multicentre cohort study},
journal = {PLOS Digital Health},
year = {2026},
volume = {5},
number = {9},
pages = {e0001735},
doi = {10.1371/journal.pdig.0001735}
}
EUVIMED