2024 World Conference on Lung Cancer (WCLC) - Posters-P3.13C.06 Artificial Intelligence Meets SCLC - Integrating Clinicopathological and Whole-Slide Image Data for Prognostic Prediction in SCLC.

P3.13C.06 Artificial Intelligence Meets SCLC - Integrating Clinicopathological and Whole-Slide Image Data for Prognostic Prediction in SCLC.

Back to course

Pdf Summary

This study explores the use of artificial intelligence (AI), specifically a deep learning (DL) model, to predict survival outcomes in patients with small-cell lung cancer (SCLC). SCLC is known for its aggressive behavior and poor prognosis, making early and accurate prognostic predictions crucial for devising personalized treatment plans.<br /><br />The study comprised 307 SCLC patients, divided into a training cohort of 263 patients treated at Hospital del Mar in Barcelona, and a validation cohort of 44 patients from the CANTABRICO phase IIIB clinical trial. Researchers built a random forest (RF) model using clinical data and a DL model based on whole-slide imaging (WSI) of pathology slides. The performance of these models was evaluated using the area under the receiver operating characteristic curve (AUC) through 5-fold cross-validation.<br /><br />Results showed that the RF model achieved an AUC of 0.728 in predicting long-term overall survival (LT_OS) in the training cohort, while the combined RF and DL model slightly improved the prediction with an AUC of 0.744. For long-term progression-free survival (LT_PFS), the RF model had an AUC of 0.689, and the combined model achieved 0.704. In the validation cohort, the combined model yielded an AUC of 0.604 for LT_OS and 0.690 for LT_PFS.<br /><br />The study concludes that integrating clinicopathological data with advanced digital pathology models could improve outcome predictions in SCLC patients. It suggests that adding more biomarkers, such as immunohistochemical markers or genomic alterations, could enhance the predictive power of such AI models, ultimately aiding in personalized patient care and treatment strategies.

Asset Subtitle

Pedro Rocha

Meta Tag

Speaker Pedro Rocha

Topic SCLC & Neuroendocrine Tumors

Keywords

artificial intelligence

deep learning

small-cell lung cancer

prognostic predictions

random forest model

whole-slide imaging

survival outcomes

clinicopathological data

digital pathology

personalized treatment