Recommendations from The Medical Education Editor.

idiopathic interstitial pneu-monia or connective tissue disease-related ILD. Median survival was just over 7 years. Accuracy, sensitivity, specific-ity, positive predictive value and negative predictive value for diagnosing IPF by the algorithm were all above 80%. Kaplan – Meier curves for survival were similar between algorithm and MDD diagnosis groups. In the group of patients diagnosed as IPF by MDD, the patients the algorithm also diagnosed as IPF had significantly worse survival than those it designated non-IPF. The results from Furukawa et al. demonstrate how machine learning algorithms might augment human diagnosis in the future. Of course, their algorithm needs testing in other cohorts to confirm these findings, and other groups are working on similar models. 7 declared.