Research questionHow can multimodal models predict lung cancer survival when imaging, clinical, and genomic data are inconsistently missing over time?Clinical datasets combine imaging, records, genomic measurements, and follow-up collected at different times, with some modalities absent for particular patients. This makes it difficult to determine whether survival models can use complementary evidence without relying on complete, consistently observed inputs. Latest papersRecent research connected to this question, newest first.Real-World Multi-Modal and Longitudinal Lung Cancer DatasetThe evidence concerns a 1,365-patient, multicenter lung cancer dataset containing whole-slide images, CT and PET scans, structured clinical data, transcriptomics, and longitudinal treatment and follow-up information. It supports uni-modal and multi-modal benchmarks for 12-month overall survival, disease-specific survival, and longitudinal hazard prediction under severe missing data; each imaging modality has multiple instances.research paper · Sep 4, 2026