Automated HIV Case Identification from the MIMIC-IV Database.

Area of Science:

• Medical Informatics
• Epidemiology
• Public Health

Background:

• Electronic Health Records (EHRs) are crucial for medical research, particularly for studying diseases like Human Immunodeficiency Virus (HIV).
• The MIMIC-IV database, an extension of MIMIC-III, offers extensive hospital admission data valuable for secondary research.
• Existing HIV phenotyping algorithms lack comprehensive application to MIMIC-IV and do not incorporate newer HIV-1/HIV-2 antibody differentiation immunoassay tests.

Purpose of the Study:

• To provide insights into the structure and data elements within the MIMIC-IV database relevant to HIV research.
• To develop and propose a novel HIV phenotyping algorithm tailored for the MIMIC-IV database.
• To address the gap in HIV case identification within MIMIC-IV and incorporate advanced diagnostic tests.

Main Methods:

• Exploration and analysis of MIMIC-IV database structure and relevant data tables.
• Development of a new phenotyping algorithm to accurately identify HIV cases.
• Inclusion of criteria for HIV-1/HIV-2 antibody differentiation immunoassay tests in the algorithm.

Main Results:

• Identification of specific data tables and elements within MIMIC-IV utilized for HIV phenotyping.
• Extraction of 1,781 HIV cases from MIMIC-IV version 0.4 and 1,843 cases from version 2.1.
• Provision of summary statistics for the identified HIV case cohorts.

Conclusions:

• The developed HIV phenotyping algorithm effectively identifies HIV cases in MIMIC-IV, incorporating newer diagnostic methods.
• The identified HIV case cohorts and associated statistics serve as a foundation for future statistical and machine learning model development.
• This work facilitates advanced research on HIV using large-scale EHR data.