Abstract by Alexander Fabiano
Automated Discovery of Diagnostic and Procedural Episodes of Care
3M (3 minute presentation)
The aim of our research is to identify patterns of care corresponding ICD-9 codes in order to predict patient health profiles. The automated discovery of healthcare episodes could add substantial value to current medical processes. With the advent of EHR, large amounts patient medical data have become available for analysis. For part one of our research, we plan to create low-density matrices using HCUP data to identify patterns we will call, “episodes of care,” our term for specific instances of treatment. These episodes will correspond characteristic groups and allow for more accurate healthcare profile classifications and predictions. For part two, we will use these procedure and diagnosis probabilities to predict an individual patient’s likelihood of temporal transitioning from one diagnosis or procedure to another.