Using State Data to Predict a Single Institution Mortality for Patients That Fall


For patients aged 45 years and older, falls are the leading cause of injury death. A recent study featuring PISC Senior Scholar Elinore Kaufman, MD, MSHP, and Past Trainee James Byrne, MD, PhD, evaluated whether a machine learning algorithm could accurately predict injury outcomes. To create an algorithm, the team collected data from Pennsylvanian fall injury patients in 2009-2019. Prediction modeling includes thirteen variables available upon patient arrival. Results indicate that 180,284 patients met the inclusion criteria, with an average mortality rate of 4.0%. The AUC for predicting patient mortality fell between 0.797 and 0.880. The team suggests the machine learning algorithm can be implemented by many local trauma systems throughout Pennsylvania. Fall mortality prediction systems can help healthcare providers navigate resource allocation among patients, as well as direct healthcare providers to patients that require the most care.


Go to Article

Back to News