WCLC 2025 - Posters & ePosters-P1.07.22 LANTERN1: AI Model for Postoperative Complications Prediction for NSCLC Lung Resection: Prospective Multicentric Study and Extern Validation

P1.07.22 LANTERN1: AI Model for Postoperative Complications Prediction for NSCLC Lung Resection: Prospective Multicentric Study and Extern Validation

Back to course

Pdf Summary

The LANTERN-1 study is a multicenter, multiomics observational clinical trial across five European institutions aimed at improving lung cancer management by integrating traditional clinical data with advanced AI techniques to predict post-operative complications in non-small cell lung cancer (NSCLC) patients undergoing surgery. Between May 2023 and March 2024, 212 patients were prospectively enrolled, with 170 undergoing surgery and 42 excluded due to advanced stage disease.<br /><br />Researchers collected extensive data including 59 clinical characteristics and 52 spirometry variables. They applied feature selection methods such as Mutual Information Maximization (MIM), Minimum Redundancy Maximum Relevance (MRMR), and Lasso logistic regression. Predictive models included logistic regression, elastic-net logistic regression, and random forest, evaluated through 3-fold cross-validation using the Area Under the Curve (AUC) metric.<br /><br />Among patients (mean age 67.6 years), common surgeries were lobectomy (57.8%), sub-lobar resections (25.9%), and a minority of bilobectomy and pneumonectomy. Post-operative complications occurred in 17.1% of patients, graded by Clavien-Dindo scale. Key predictors identified were the Charlson comorbidity index (AUC 0.70) and preoperative forced expiratory volume in 1 second (FEV1) (AUC 0.77). Combining these variables enhanced predictive performance to an AUC of 0.82, indicating a strong model for anticipating complications.<br /><br />Clinically, the AI-driven model offers valuable support for patient counseling, surgical decision-making, identifying high-risk patients, and optimizing resource allocation. The LANTERN-1 study demonstrates the potential of integrating comprehensive clinical data with AI to improve outcome predictions in NSCLC surgical care, aiming to personalize treatment plans and reduce post-operative risks.

Asset Subtitle

Filippo Lococo

Meta Tag

Speaker Filippo Lococo

Topic Early-Stage Non-small Cell Lung Cancer

Keywords

LANTERN-1 study

non-small cell lung cancer

NSCLC

post-operative complications

AI predictive model

feature selection

logistic regression

random forest

Charlson comorbidity index

forced expiratory volume (FEV1)