Deep Learning for predicting urgent hospitalizations
in elderly population using administrative EHR
DL on administrative EHR
Administrative Electronic Health Records (EHR) collect a big amount of data, including demographic information, diagnoses recorded during hospitalizations, medication prescriptions, and more. This information reflects the disease progression of each subject. Although EHRs were born to improve the efficiency of health systems, today they are extensively used to epidemiological studies. In this project, Deep Learning is applied to learn the history of polypharmacy and multimorbidity of elderly population (>65 years) in Piedmont. Medication prescriptions and hospitalization diagnoses from administrative EHRs are used to reconstruct their health trajectories, for predicting future urgent hospitalizations.
Veronica Sciannameo e Paola Berchialla