Clinical usability of an explainable AI decision support tool and evaluation of multimodal models in NSCLC

A multimodal explainable AI model outperformed established biomarkers such as PD-L1 expression in predicting immunotherapy outcomes for patients with non-small cell lung cancer in a large international real-world study. The tool integrated multiple data modalities and provided interpretable outputs that improved physician decision-making. The study, published in Nature Medicine, represents one of the largest clinical validations of an AI decision-support system in thoracic oncology to date.