Predicting a failure of postoperative thromboprophylaxis in non-small cell lung cancer: A stacking machine learning approach
PLOS ONE · 2025
DOI: 10.1371/journal.pone.0320674 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS ONE, 2025. Verfasst von Ligang Hao und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Thoracic Surgery, Xingtai People's Hospital, Xingtai, Hebei, China. Hauptthemen laut MEDLINE: Carcinoma, Non-Small-Cell Lung, Machine Learning, Lung Neoplasms, Venous Thromboembolism. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: the Key Development Plan of XingTai.
Themen: Thromboseprophylaxe, Künstliche Intelligenz in der Pflege
Abstract (Originalsprache)
Background
Non-small-cell lung cancer (NSCLC) and its surgery significantly increase the venous thromboembolism (VTE) risk. This study explored the VTE risk factors and established a machine-learning model to predict a failure of postoperative thromboprophylaxis.
Methods
This retrospective study included patients with NSCLC who underwent surgery between January 2018 and November 2022. The patients were randomized 7:3 to the training and test sets. Nine machine learning models were constructed. The three most predictive machine-learning classifiers were chosen as the first layer of the stacking machine-learning model, and logistic regression was the second layer of the meta-learning model.
Results
This study included 362 patients, including 58 (16.0%) with VTE. Based on the multivariable logistic regression analysis, age, platelets, D-dimers, albumin, smoking history, and epidermal growth factor receptor (EGFR) exon 21 mutation were used to develop the nine machine-learning models. LGBM Classifier, RandomForest Classifier, and GNB were chosen for the first layer of the stacking machine learning model. The area under the received operating characteristics curve (ROC-AUC), accuracy, sensitivity, and specificity of the stacking machine learning model in the training/test set were 0.984/0.979, 0.949/0.954, 0.935/1.000, and 0.958/0.887, respectively. In the validation set, the final stacking machine learning model demonstrated an ROC AUC of 0.983, accuracy of 0.937, sensitivity of 0.978, and specificity of 0.947. The decision curve analyses revealed high benefits.
Conclusion
The stacking machine learning model based on EGFR mutation and clinical characteristics had a predictive value for postoperative VTE in patients with NSCLC.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- Department of Thoracic Surgery, Xingtai People's Hospital, Xingtai, Hebei, China.
- Department of Computed Tomography and Magnetic Resonance, Xingtai People's Hospital, Xingtai, Hebei, China.
- Department of Clinical Lab, Xingtai People's Hospital, Xingtai, Hebei, China.
ORCID-Profile: Yonghui Di
Förderung
- the Key Development Plan of XingTai
Zitieren
APA
Hao, L., Zhang, J., Di, Y., Qi, Z., & Zhang, P. (2025). Predicting a failure of postoperative thromboprophylaxis in non-small cell lung cancer: A stacking machine learning approach. PLOS ONE. https://doi.org/10.1371/journal.pone.0320674
Vancouver
Hao L, Zhang J, Di Y, Qi Z, Zhang P. Predicting a failure of postoperative thromboprophylaxis in non-small cell lung cancer: A stacking machine learning approach. PLOS ONE. 2025. doi:10.1371/journal.pone.0320674
BibTeX
@article{Hao2025,
author = {Hao, Ligang and Zhang, Junjie and Di, Yonghui and Qi, Zheng and Zhang, Peng},
title = {Predicting a failure of postoperative thromboprophylaxis in non-small cell lung cancer: A stacking machine learning approach},
journal = {PLOS ONE},
year = {2025},
doi = {10.1371/journal.pone.0320674}
}
EUVIMED