2024 World Conference on Lung Cancer (WCLC) - Posters-P1.06A.01 Machine Learning Models Discriminate Independent Primaries with Metastatic in Non-Small-Cell Lung Cancer

P1.06A.01 Machine Learning Models Discriminate Independent Primaries with Metastatic in Non-Small-Cell Lung Cancer

Pdf Summary

The document outlines a study on the application of machine learning (ML) models for discriminating between independent primary multiple lung cancers (MPLC) and intrapulmonary metastases (IPM) in non-small-cell lung cancers (NSCLCs). The prevalence of multiple NSCLCs has risen due to advancements in lung cancer screening, emphasizing the need for accurate discrimination between MPLC and IPM. This differentiation is vital for the appropriate treatment and management of the disease but remains challenging.<br /><br />The study explores the potential of various ML models, like random forests (RF), decision trees (DT), and gradient boosting decision trees (GBDT), at the molecular level, particularly focusing on whole-exome sequencing (WES) data. These models demonstrate high accuracy in the discrimination of MPLC from IPM. The effective use of these algorithms can streamline classification processes, assisting clinicians in making informed decisions about patient treatment strategies.<br /><br />The results show that ML models not only offer precise discrimination based on molecular profiles but also highlight differences in disease-free survival (DFS) times among undetermined multiple NSCLCs, providing further insight into patient prognosis.<br /><br />Conclusively, the study suggests that ML models can significantly aid clinicians by providing robust, one-step classifications, especially in complex clinical cases. The research stresses the need for additional multicenter studies to validate and extend these findings, potentially leading to broader implementation in clinical practice.<br /><br />The study represents a promising advancement in the medical field, exhibiting how contemporary ML techniques can enhance the accuracy and efficiency of lung cancer diagnosis and treatment.

Asset Subtitle

Ning Liu

Meta Tag

Speaker Ning Liu

Topic Pathology & Biomarkers

Keywords

machine learning

multiple lung cancers

intrapulmonary metastases

non-small-cell lung cancer

random forests

decision trees

gradient boosting

whole-exome sequencing

disease-free survival

clinical practice