Medicine is a data science, we should teach like it
Medicine has always been a data science. Collecting and interpreting data is a key component of every interaction between physicians and patients. Data can be anything from blood pressure measurements at a yearly exam to complex radiology images interpreted by experts or algorithms. Interpreting these uncertain data for accurate diagnosis, management, and care is a critical component of every physician’s daily life. The intimate relationship between data science and medicine is apparent in the pages of our most prominent medical journals. Using Pubmed, we pulled the abstracts of all papers published in The New England Journal of Medicine, JAMA, Nature Medicine, The Lancet, PLoS Medicine, and BMJ for the years 2010 - March 2019. We then searched for a list of statistical terms in the text of these abstracts. For these 12,281 abstracts a median of 50% (IQR 30%, 67%) of sentences contained a term that would require statistical training to understand.