Medical errors are the third leading cause of death in the U.S., accounting for 10 percent of all deaths, according to Johns Hopkins’ researchers. To improve patient safety in its four large metropolitan hospitals and other facilities, Allina Health turned to data analytics to standardize and expand safety event reporting. This presentation will review the efficacy of machine learning to identify and implement triggers that predict when a patient may be harmed or indicate if harm may have occurred. Presenters will share findings from an 11-month pilot finding, that the trigger tool turned up 333 pressure injuries that were not in the voluntary reporting system, enabling a wound ostomy nurse to focus on reversing the course of injury much earlier than otherwise. A retrospective comparison of the predictive risk model with the standard Braden score found that the model more accurately identified pressure injury patients and reduced the number of patients misclassified as being at risk.