Multi-omics Model for Pancreatic Cancer Screening Using cfDNA

Peripheral blood samples from participants with new diagnosis of pancreatic cancers, precancerous lesions and healthy individuals will be collected to characterize the cancer-specific signatures by low-pass whole-genome sequencing and target methylation sequencing using cfDNA. An ensemble multi-omics model will be trained aided by machine learning algorithm and validated in test set. The performance of this multi-omics model distinguishing pancreatic cancer from non-cancer will be evaluated.