The landscape of conventional and artificial intelligence-based clinical prediction models in non-small-cell lung cancer: from development to real-world validation.

Globally, lung cancer remains the most common cause of cancer mortality, with non-small-cell lung cancer (NSCLC) being the most common subtype of lung cancer diagnosed. This review paper provides a comprehensive landscape of clinical prediction models (CPMs) in NSCLC, including in early-stage and metastatic disease, and the recent acceleration of artificial intelligence integration. Prediction models are developed using multimodal patient data to allow oncologists to make evidence-based decisions regarding patient treatment options. Despite these models in early-stage and metastatic NSCLC showing promise, their clinical application provides challenges, involving an unmet need for external validation, alongside a lack of prospective modelling. However, the continued advancements in this field, comprising production and accessibility of large-scale pathology databases and external validation of developed models, allow for continued research and progress. These models have potential to assist in personalised treatment selection, supporting oncologists in perceiving future risk factors or issues associated with a specific targeted therapy for an individual patient, ultimately optimising treatment to precise, personalised options for individuals diagnosed with NSCLC.