Harnessing machine learning and AI-driven analytics to identify novel drug targets and predict chemotherapy efficacy in NSCLC.

Introduction: Non-small cell lung cancer (NSCLC) constitutes the majority of lung cancer cases and exhibits marked heterogeneity in both clinical presentation and molecular profiles, leading to variable responses to chemotherapy. Emerging evidence suggests that mitochondria-derived RNAs (mtRNAs) may serve as novel biomarkers, although their role in predicting chemotherapy outcomes remains to be fully explored.

Methods: In this study, peripheral blood mononuclear cells were obtained from NSCLC patients for analysis of the mtRNA ratio (mt_tRNA-Tyr-GTA_5_end to mt_tRNA-Phe-GAA), while thoracic CT images were processed to derive an AI-driven BiomedGPT variable. Although individual clinical factors (Sex, Age, History_of_smoking, Pathological_type, Stage) offered limited predictive power when used in isolation, their integration into a random forest model improved sensitivity in the training set, albeit with reduced generalizability in the validation cohort. The subsequent integration of the BiomedGPT score and mtRNA ratio significantly enhanced predictive performance across both training and validation datasets.

Results: An all-inclusive model combining clinical data, AI-derived variables, and mtRNA biomarkers produced a risk score capable of discriminating patients into high- and low-risk groups for progression-free survival and overall survival, with statistically significant differences observed between these groups.

Discussion: These findings highlight the potential of integrating mtRNA biomarkers with advanced AI methods to refine therapeutic decision-making in NSCLC, underscoring the importance of combining diverse data sources in precision oncology.