Abstract: Escalating healthcare costs and inconsistent quality is exacerbated by clinical practice variability. Diagnostic testing is the highest volume medical activity, but human intuition is typically unreliable for quantitative inferences on diagnostic performance characteristics. Electronic medical records from a tertiary academic hospital (2008-2014) allow us to systematically predict laboratory pre-test probabilities of being normal under different conditions. We find that low yield laboratory tests are common (e.g., ~90% of blood cultures are normal). Clinical decision support could triage cases based on available data, such as consecutive use (e.g., lactate, potassium, and troponin are >90% normal given two previously normal results) or more complex patterns assimilated through common machine learning methods (nearly 100% precision for the top 1% of several example labs).
Learning Objective 1: Explain how predictive algorithms can estimate the distribution of pre-test probability for various laboratory tests and how that can be used to identify low yield tests.
Learning Objective 2 (Optional): Name common lab tests with pre-test probability of being normal >90% given prior consecutive normal results.
Shivaal Roy (Presenter)
Jason Hom, Stanford University
Lester Mackey, Microsoft Research New England
Neil Shah, Stanford University
Jonathan Chen, Stanford University