__STYLES__
Analyze and build reports around patient, encounter, and procedure data to provide KPI and operational performance reports, and to uncover deeper insights from the dataset.
1. Data Exploration
Goals achieved in this step:
2. Data Preparation
Goal: Prepare the data for an effective data model for analysis.
Key steps taken:
3. Data Model
With the previous steps and preparations, the data model can be established for dashboard building.
Special consideration was given to the procedures table. Although there are possible links to dimension tables, the only relationship set is to the encounter table to ensure the correct linkage between encounters and procedures.
4. Logical Considerations While Building the Dashboard
Two major considerations were made when building the logic into the analysis:
The dashboard is using fiscal year as the years range instead of calendar year.
By definition, only the Inpatient class includes overnight visits, which should be identified as admissions. However, data showed that visits under different encounter classes also had overnight stays, with some showing hundreds of days of stay. Thus, admissions and non-admissions were built by looking at the days stayed in the hospital rather than only the encounter classes.
For readmission rate calculation, initial admissions and readmissions need to be identified. Initially, only these two types were considered based on visit dates. However, many visits were regular daily visits over consecutive days and should not be counted as initial admissions or readmissions. After careful review, these were categorized as a single admission for readmission rate calculation, while still counted as individual visits in the general admission settings.