Machine Learning Model for Predicting Severe Adverse Events in Oncology Patients Using the US Food and Drug Administration Adverse Event Reporting System
Machine-Learning-Modell zur Vorhersage schwerer unerwünschter Ereignisse bei onkologischen Patient:innen mit dem US Food and Drug Administration Adverse Event Reporting System
JCO Clinical Cancer Informatics · 2026-01 · 10 (1)
DOI: 10.1200/cci-25-00081 ↗Auf einen Blick
Ein LightGBM-Modell sagte schwere onkologische unerwünschte Ereignisse anhand von FAERS-Daten besser voraus als die logistische Regression. Wichtige Prädiktoren waren Alter, Polypharmazie und Therapiedauer.
Fragestellung
Lässt sich mit FAERS-Daten ein ML-Modell entwickeln und validieren, das schwere onkologische unerwünschte Ereignisse vorhersagt und Risikofaktoren identifiziert?
Methode
Analyse von 3,789,273 eindeutigen onkologischen FAERS-Fällen (2012Q4-2024Q3) nach Preprocessing; LightGBM mit Optuna-Optimierung, Vergleich mit logistischer Regression; Bewertung mit AUROC, AUPRC, Precision, Recall, F1-Score und SHAP.
Ergebnisse
LightGBM übertraf die logistische Regression: AUROC 0,806 (95 %-KI 0,804–0,807) vs. 0,708 (0,706–0,709); AUPRC 0,615 (0,611–0,617) vs. 0,454 (0,449–0,455); F1 78,1 % vs. 71,6 %.
Schlussfolgerung der Autor:innen
Das Gradient-Boosting-Modell verbesserte die Vorhersage und Interpretierbarkeit schwerer onkologischer unerwünschter Ereignisse gegenüber der logistischen Regression.
Kernaussagen
- AUROC des LightGBM-Modells: 0,806 (95 %-KI 0,804–0,807)
- Wichtige Prädiktoren: Alter, Polypharmazie, Therapiedauer
- SHAP zeigte eine gemeinsame Zunahme des SAE-Risikos durch diese Faktoren
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 JCO Clinical Cancer Informatics, 2026-01. Verfasst von Luke Xiyu Zhao und 3 weiteren Autor:innen. Beteiligte Einrichtung: Johns Hopkins University School of Medicine, Baltimore, MD. Hauptthemen laut MEDLINE: Adverse Drug Reaction Reporting Systems, Machine Learning, Drug-Related Side Effects and Adverse Reactions, Neoplasms. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 28 Literaturangaben.
Themen: Künstliche Intelligenz in der Pflege, Patientensicherheit
Abstract (Originalsprache)
PURPOSE Predicting severe adverse events (SAEs) in oncology is challenging because of complex therapies and patient heterogeneity. Traditional pharmacovigilance methods often fail to capture multifactorial risk patterns. Machine learning (ML) offers potential to identify subtle predictors of SAEs within large real-world data sets such as the US Food and Drug Administration Adverse Event Reporting System (FAERS). This study developed and validated an ML model to predict severe oncology-related adverse events and identify key risk factors using FAERS data. METHODS We analyzed 3,789,273 unique oncology-related FAERS cases (2012Q4-2024Q3) after extensive preprocessing, including natural language processing-based indication filtering, deduplication, and variable standardization. Severe events were defined by outcomes of death, hospitalization, disability, congenital anomaly, or life-threatening condition. A LightGBM model was trained using Optuna-based hyperparameter optimization and benchmarked against logistic regression. Model performance was evaluated using AUROC, AUPRC, precision, recall, and F1-score. Shapley Additive Explanations (SHAP) analysis assessed the feature influence and interpretability. RESULTS LightGBM outperformed logistic regression (AUROC, 0.806 [95% CI, 0.804 to 0.807] v 0.708 [0.706 to 0.709]; AUPRC, 0.615 [0.611 to 0.617] v 0.454 [0.449 to 0.455]; F1 78.1% v 71.6%). Key predictors of severity included advanced age, higher weight, extensive polypharmacy (median 15 drugs; IQR, 9-27), longer therapy duration (median 6.3 days), and greater numbers of reported reactions (mean 5 per case). SHAP analysis revealed that age, polypharmacy, and therapy duration synergistically increased SAE risk. CONCLUSION Our gradient boosting model substantially improved prediction and interpretability of severe oncology adverse events compared with logistic regression. SHAP analysis identified clinically meaningful predictors, enabling precision pharmacovigilance and targeted risk mitigation. These findings support ML integration into regulatory and clinical pharmacovigilance workflows to enhance postmarket safety surveillance.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- Johns Hopkins University School of Medicine, Baltimore, MD.
- University of Rochester School of Medicine and Dentistry, Rochester, NY.
ORCID-Profile: Luke Xiyu Zhao, Catherine Wang
Zitieren
APA
Zhao, L. X., Wang, C., Zou, J., & Ranasinghe, P. (2026). Machine Learning Model for Predicting Severe Adverse Events in Oncology Patients Using the US Food and Drug Administration Adverse Event Reporting System. JCO Clinical Cancer Informatics, 10(1). https://doi.org/10.1200/cci-25-00081
Vancouver
Zhao LX, Wang C, Zou J, Ranasinghe P. Machine Learning Model for Predicting Severe Adverse Events in Oncology Patients Using the US Food and Drug Administration Adverse Event Reporting System. JCO Clinical Cancer Informatics. 2026;10(1). doi:10.1200/cci-25-00081
BibTeX
@article{Zhao2026,
author = {Zhao, Luke Xiyu and Wang, Catherine and Zou, Jonathan and Ranasinghe, Padmini},
title = {Machine Learning Model for Predicting Severe Adverse Events in Oncology Patients Using the US Food and Drug Administration Adverse Event Reporting System},
journal = {JCO Clinical Cancer Informatics},
year = {2026},
volume = {10},
number = {1},
doi = {10.1200/cci-25-00081}
}
EUVIMED