Metastasis is the most important factor correlated with poor prognosis in cancer patients and a major driver of treatment and therapy. But with low fill rates for metastasis and site of metastasis in structured data and with manual curation being time-consuming and less scalable for extracting this valuable information from unstructured notes, there is a need for automated methods. Principal AI Scientist Krishna Swaminathan presents results and implications for industry from an ASCO 2020 study abstract which sought to validate Natural Language Processing (NLP) algorithms that accurately impute metastatic status and site of metastasis from unstructured notes.