Prognostic prediction of pulmonary neuroendocrine carcinoma using a transformer-based artificial intelligence model: a two-center retrospective cohort study
- 1Department of Thoracic Surgery, Peking Union Medical College Hospital, Chinese Academy of Medical Science & Peking Union Medical College
- 2Department of Thoracic Surgery, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College
- 3Institute of Advanced Research, Infervision Medical Technology Co., Ltd
Abstract
Background: Pulmonary neuroendocrine carcinomas (pNECs) are a rare and heterogeneous group of malignancies, ranging from low-grade carcinoids to high-grade small-cell lung carcinoma (SCLC) and large-cell neuroendocrine carcinoma (LCNEC). Prognostic tools for these tumors are scarce and have rarely been validated in external cohorts.
Objectives: This study aimed to develop a Transformer-based prognostic prediction model for pNEC and validate it in an independent external cohort, to improve survival estimation and support individualized treatment decisions.
Methods: This retrospective two-center cohort study included 823 patients with surgically resected pNEC (2001-2023). Models were developed in Center 1 (n=374; randomly split 8:2 into training and internal test sets) and externally validated in an independent Center 2 cohort (n=449). Three algorithms—logistic regression (LR), decision tree (DT), and the Transformer-based tabular foundation model TabPFN—were compared for estimating overall survival (OS) and disease-free survival (DFS) at 1-, 3-, and 5-year landmarks, with evaluation by receiver operating characteristic, calibration, and decision curve analyses, and SHapley Additive exPlanations (SHAP) analysis for interpretability.
Results: TabPFN achieved the highest area under the curve (AUC) in most tasks in both test sets, with task-dependent exceptions: in the internal test set, logistic regression performed best for 1-year OS (AUC 0.885), while TabPFN performed best for the remaining tasks, including 3-year OS (0.880), overall DFS (0.810), and 1-year DFS (0.892); in the external test set, TabPFN remained best for 1- and 3-year OS and for overall, 1-, and 3-year DFS (0.747-0.802), whereas logistic regression performed best for 5-year OS and DFS. External calibration was less satisfactory. Disease stage, pathological subtype, and age were the key prognostic predictors, and TabPFN was more accurate for low-grade subtypes. SHAP analysis quantified the contribution of each feature and made the model interpretable.
Conclusions: The Transformer-based model improves survival estimation in pNEC, with task-dependent advantages over conventional methods and confirmed transportability to an independent institutional cohort. Prospective multicenter validation and calibration refinement are needed before broader clinical use.
Related
- MRI-derived habitat radiomics integrated with transformer deep learning for preoperative prediction of high CD9 expression: a novel prognostic marker in adult-type diffuse gliomas
- Artificial intelligence in hepatitis: Advances in diagnosis, treatment optimization, and prognostic prediction
- From Splicing to Survival: Understanding the Molecular Basis of Prognostic Differences in advanced Small Intestine Neuroendocrine Tumors
- Neutrophil-to-Lymphocyte Ratio as a Prognostic Biomarker in Pancreatic Neuroendocrine Neoplasms
- Prognostic factors and survival outcomes of multimodal treatment in sinonasal neuroendocrine tumors in a population based SEER analysis
BibTeX
@article{liu2026prognostic,
title = {{Prognostic prediction of pulmonary neuroendocrine carcinoma using a transformer-based artificial intelligence model: a two-center retrospective cohort study}},
author = {Lei Liu and Linchuan Liang and Dawei Wang and Jiaqi Zhang and Ke Zhao and Chao Guo and Cheng Huang and Shanqing Li and Yushun Gao and Yeye Chen},
journal = {Frontiers in Oncology},
year = {2026},
month = oct,
doi = {10.3389/fonc.2026.1948262},
url = {https://www.frontiersin.org/articles/10.3389/fonc.2026.1948262/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Prognostic prediction of pulmonary neuroendocrine carcinoma using a transformer-based artificial intelligence model: a two-center retrospective cohort study
AU - Lei Liu
AU - Linchuan Liang
AU - Dawei Wang
AU - Jiaqi Zhang
AU - Ke Zhao
AU - Chao Guo
AU - Cheng Huang
AU - Shanqing Li
AU - Yushun Gao
AU - Yeye Chen
PY - 2026
DA - 2026/10/23
JO - Frontiers in Oncology
DO - 10.3389/fonc.2026.1948262
UR - https://www.frontiersin.org/articles/10.3389/fonc.2026.1948262/full
AB - Background: Pulmonary neuroendocrine carcinomas (pNECs) are a rare and heterogeneous group of malignancies, ranging from low-grade carcinoids to high-grade small-cell lung carcinoma (SCLC) and large-cell neuroendocrine carcinoma (LCNEC). Prognostic tools for these tumors are scarce and have rarely been validated in external cohorts. Objectives: This study aimed to develop a Transformer-based prognostic prediction model for pNEC and validate it in an independent external cohort, to improve survival estimation and support individualized treatment decisions. Methods: This retrospective two-center cohort study included 823 patients with surgically resected pNEC (2001-2023). Models were developed in Center 1 (n=374; randomly split 8:2 into training and internal test sets) and externally validated in an independent Center 2 cohort (n=449). Three algorithms—logistic regression (LR), decision tree (DT), and the Transformer-based tabular foundation model TabPFN—were compared for estimating overall survival (OS) and disease-free survival (DFS) at 1-, 3-, and 5-year landmarks, with evaluation by receiver operating characteristic, calibration, and decision curve analyses, and SHapley Additive exPlanations (SHAP) analysis for interpretability. Results: TabPFN achieved the highest area under the curve (AUC) in most tasks in both test sets, with task-dependent exceptions: in the internal test set, logistic regression performed best for 1-year OS (AUC 0.885), while TabPFN performed best for the remaining tasks, including 3-year OS (0.880), overall DFS (0.810), and 1-year DFS (0.892); in the external test set, TabPFN remained best for 1- and 3-year OS and for overall, 1-, and 3-year DFS (0.747-0.802), whereas logistic regression performed best for 5-year OS and DFS. External calibration was less satisfactory. Disease stage, pathological subtype, and age were the key prognostic predictors, and TabPFN was more accurate for low-grade subtypes. SHAP analysis quantified the contribution of each feature and made the model interpretable. Conclusions: The Transformer-based model improves survival estimation in pNEC, with task-dependent advantages over conventional methods and confirmed transportability to an independent institutional cohort. Prospective multicenter validation and calibration refinement are needed before broader clinical use.
SN - 2234-943X
ER -