Figure 1.

Study overview.

(A) Detection of microsatellite instability by deep learning could be used as a screening tool and rule out a part of the patients from further testing. (B) A deep learning classifier was trained on eight out of nine cohorts and validated in the remaining one. Circle size indicates the patient number of each cohort as shown. Icon source: Twitter Twemoji (CC-BY 4.0).

AI, artificial intelligence; dMMR, mismatch repair deficiency; DACHS, Darmkrebs: Chancen der Verhütung durch Screening; MECC, Molecular Epidemiology of Colorectal Cancer; MSI, microsatellite instability; NLCS, Netherlands Cohort Study; QUASAR, Quick and Simple and Reliable; TCGA, The Cancer Genome Atlas; UMM, University Medical Center Mannheim.