Deep neural models for chronic disease status detection in free text clinical records

Abstract

Timely identification or prediction of a disease status can be a boon to patient's life. The elements that can detect the diseases status are mostly present in free text clinical records; these records contain private information about the patients. The time-consuming manual process to identify the disease status from these clinical records can be error-prone and comes with an expense. Hence the need to automate or semi automate this process is felt in the community. In this paper, we have used deep learning techniques on the publically available i2b2 clinical datasets to detect the chronic disease status, leading to promising results.