Abstract by Adam Ott
Improving Ambulatory Care Sensitive Emergency Department Admissions Predictions by Accounting for Spatial and Familial Correlation Structures
Purpose: We wish to improve the prediction of the risk of avoidable emergency department admissions due to Ambulatory Care Sensitive Conditions by incorporating familial, temporal, and spatial correlation structures alongside other risk-factors of avoidable admissions, using data provided by Intermountain Healthcare.
Methods: This project will use various correlation structures to predict unnecessary emergency department admissions. Explanatory variables will include ethnicity, gender, age, weight, marital status, religiosity, socioeconomic status, medical history, and health insurance coverage. We will split the data into a train and test set to validate any model we fit.
Results: At this point of the project, we are still in the gathering and cleaning data stage. At the end of this project, the explanatory variables and correlation structures used will help predict those that are at risk of an avoidable admission. Using the results of this study, Intermountain Healthcare should be able to reach out to patients with high risk, thus lowering the patients’ health care expenditures and improving their experience in the health care system.