Deep Learning Models for Electronic Health Record Data Analysis
Wasswa Shafik
Abstract
Electronic health records (EHRs) are digital versions of patients’ medical charts stored in shared databases within patient care settings. EHRs contain patients’ medical and treatment histories including diagnoses, medications, treatment plans, immunization dates, allergies, radiology images, and laboratory test results. EHR systems aid in the easy management of patient data, disease surveillance, drug utilization review, and democratized access to health records. Moreover, EHRs are the first source of data to support institutions’ administrative needs, such as budgeting and billing functions, and are valuable resources for evidence-based information pertinent to patient care. However, challenges with EHR data analysis include poor-quality data owing to failed checks; incomplete records; the need to analyze unstructured free text; incompatibility across different EHR systems; the need for clear data governance and use consent; and finally, the issue of patient privacy. However, successful EHR data analysis creates enormous opportunities for assessing the quality and safety of healthcare; examining disease history and prognosis; comparing different treatments; and measuring disease burden within populations. In the process, deep learning has been shown to be an effective instrument because it allows for making sense of sparse data.