MHDR Data

The MHDR utilizes the i2b2 data structure to store the clinical data for the patients of Boston Medical Center and our affiliated community health centers. The following information will provide researchers with a reference to utilize the system and understand the data in such a manner that will allow them to draw valid, reliable conclusions from the data.

What is i2b2?
i2b2 (Informatics for Integrating Biology & the Bedside) is an open source relational database system that allows researchers to access clinical medical information for a population for research or health outcomes monitoring purposes. The MHDR combines clinical medical information from the Boston Medical Center and our affiliated community health centers.

The structure of i2b2
The most important concept to understand when working within the i2b2 framework is that this system contains facts that are related to member of a population. The primary step in all research conducted in the i2b2 system is to define your population of interest – Who are you attempting to learn about?

Regardless of whether you are looking at primary care utilization rates or trends is blood pressure measurements over time, you are dealing with a population of interest. All results obtained from the i2b2 analysis or HOME cell interfaces will be presented as a measure of some facts about this population.

The basic database architecture of the MHDR i2b2 uses 3 Tables about 3 Things:

Patients

The patient_dimension table contains the demographic information about each person in the database. This includes patient number, gender, race, language, religion, marital status, zip code, date of birth.

Concepts

The concept_dimension contains the information that links the groups of concepts in the observation_fact table together through the ontologies.

Observations

The observation_fact table is the functional heart of the entire i2b2 MHDR system. All the information about the population contained in the database is stored in this table as facts. These facts are linked to the patient through the patient number and linked to time through the encounter number and the start date and end date.