Information Extraction from EMRs to Predict Readmission following Acute Myocardial Infarction

Objectives: We propose to utilize patient and disease risk factors and additional free-text notes in the medical record and geodata to improve prediction of death and/or readmission. We have three hypothesizes 1) classic risk prediction for death and/or readmissions after Acute Myocardial Infarction using patient and disease comorbidities can be improved by natural language processing techniques from free-text fields in the medical record, 2) classic risk prediction for death and/or readmissions after Acute Myocardial Infarction can be improved by geomedicine techniques by linking publically available residential risk factor data (geodata) to the patient's residential address, and 3) the development of a robust automated surveillence tool for Acute Myocardial Infarction patients at the time of discharge could improve survival and reduce readmissions.

1R01HL130828-01A1

PI: 
Wendy Chapman
Jeremiah A. Brown
Amarendra Das