Cancer outcomes predicted using AI-extracted data from clinical notes

Electronic health records and tumour-genome profiling offer promising real-world data with which to model cancer outcomes. Models built by using artificial intelligence (AI) to annotate clinical reports from more than 20,000 individuals with cancer can predict survival, as well as identify genomic features associated with metastasis and response to immunotherapy. Applying natural language processing to health records generates a data set that can be used to predict cancer outcomes.