Deep learning can predict microsatellite instability (MSI) directly from histology in gastrointestinal (GI) cancer, according to a study published in *Nature Medicine*. The research team, led by Jakob Nikolas Kather, developed a deep residual learning model that can classify MSI status from hematoxylin-eosin (HE) histology slides. This approach has the potential to identify MSI status without additional genetic or immunohistochemical tests, which are not universally available. The model, trained on large cohorts from the *The Cancer Genome Atlas* (TCGA), demonstrated high accuracy in predicting MSI status across different tumor types and ethnicities. The method was validated in external cohorts, including the DACHS study and the KCCH cohort, showing robust performance. The study highlights the potential of deep learning to enable broader MSI screening and the distribution of benefits of cancer immunotherapy to a larger patient population.Deep learning can predict microsatellite instability (MSI) directly from histology in gastrointestinal (GI) cancer, according to a study published in *Nature Medicine*. The research team, led by Jakob Nikolas Kather, developed a deep residual learning model that can classify MSI status from hematoxylin-eosin (HE) histology slides. This approach has the potential to identify MSI status without additional genetic or immunohistochemical tests, which are not universally available. The model, trained on large cohorts from the *The Cancer Genome Atlas* (TCGA), demonstrated high accuracy in predicting MSI status across different tumor types and ethnicities. The method was validated in external cohorts, including the DACHS study and the KCCH cohort, showing robust performance. The study highlights the potential of deep learning to enable broader MSI screening and the distribution of benefits of cancer immunotherapy to a larger patient population.