Data Tools | Baltimore Health Analytics

BHA Health Data Management Tools

Healthcare data is full of idiosyncrasies and our web applications are full of healthcare data.  When you engage Baltimore Health Analytics on a project, you get the benefit of our years of experience solving the challenges that these present. 

Along with that, we have a library of programs that save time by solving very specific problems that we see again and again.

Master Slider in Macbook

External file feeds where columns appear and disappear

Environments with Linux and Windows, SQL Server, Mysql and Teradata

Multiple products, spans and ids per member

Sensitive data requiring high standards for protection

Charts stored in different ways across the organization

Stars Reporting Library

The Challenge

Esthetically pleasing reports that can be used in day to day operations and executive presentations are hard to build.  You want sharp graphs that show progress towards Stars and detail that can tell at a glance how you're progressing at this point in the year.

Our Solution

Prebuilt reports that help you track rates, gaps, cutpoints, year over year progress, and the improvement measures. These are designed to be easy to interpret, to use in executive presentations and to manage your Stars improvement through HEDIS®.


The Challenge

It's really hard to create realistic test data.  But if you extract production data, you run into PHI and HIPPA concerns.

Our Solution

BHA's Extraction and Scrambling data tool uses production data as a template but creates a deidentified dataset that maintains the same referential integrity that can't be traced back to the original patient. ExScrambler is a script that unwinds the table dependencies, scrambles the patient names, birth dates and service dates and builds a copy of each table for a random sample of members.

But it's not random data and so it maintains the utility of using the patient record for development. Ruiz Gonzalez of 123 Oak St who had cancer when he was 56 becomes the template for William Ahmed, of 4245 Main St who also had cancer at age 56 but on a different date.

Member Filename Matcher

The Challenge

Your vendors, your medical management departments, your network providers -- all of your partners -- have medical charts readily available for your chart abstraction efforts. But you can't use them because they all have different systems for storing and naming the files. One system has SMITH_JOHN_09091945.pdf another has John Smith Sr 9-9-1945.pdf and another has SMITH_J_356245.pdf, but they're all the same person. You could go back and ask each partner to create a mapping file, but how long will that take?

Our Solution

We created MemberFilenameMatcher to solve this situation and make charts usable for HEDIS® abstraction. It figures out all the ways you might write a member id, date of birth and name, then applies search rules to search the filesystem for a match -- including zip files. It even gathers the files together and creates a manifest list for upload into BHA's Ocular OCR assisted abstraction system.

File Merger

The Challenge

You have dozens of files with mostly the same columns, but not quite. You can't just concatenate the files because one of them has a lab value or or other extra content. It'd be much easier to manage if you could put it all into the same table.

member_id, birth_date, member_name, address, a1c_value, a1c_dos
member_id, birth_date, member_name, address, col_date, fobt_date, sig_date
member_id, birth_date, member_name, address, bcs_date, provider_name

Our Solution

Our file merger will pick through all the files in all of your directories, read the column headers and then create a single file that has all the common fields, plus all the unique fields.

Data Mover

The Challenge

You have different files files in different formats and have different databases. You can't control the upstream source -- if they add a column or remove a column then you have to revise your script even if it's a column you don't need. That's easy enough to do if it's one source to one database, but what if you have .csv, text, and xlsx and need to go to Teradata and SQL Server?

Our Solution

DataMover takes input files and copies them to a database table.  It analyzes the source file to determine the datatypes and collect instrumentation, No additional tools are needed, it just works.  If the input file format changes, the target table matches.  Once the files are staged, we have libraries to run additional stored procedures and queries to complete the processing.

Technical Info

The BHA data tools are written as Perl modules. They are designed so that a single Perl script can handle data load through reporting on Linux or Windows platforms.