Today is the day I begin a journey to use public health data in ways to make a better world.
How can I to change the world without being just another techy dilettante with good intentions? I honestly don't know, but I believe I can. With 4 years in a top-rated Mathematics PhD program and another 20 years as a developer under my belt, I have as many tools as anyone could want. In the mathematical parlance: I'll solve one important problem that I can solve, I'll see a bigger problem, then solve that one.
What Kind of Data can I gather to help Understand the Mental Health Crisis?
In line with the current news, the ER is the primary place of treat for those in mental health crisis. It's also one of the worst places to get treatment.
Pulling data together in a GIS style is a good start. We need multiple data sets, because generally one organization can put together a good data set, but combining and comprehending data across organizational boundaries takes normalization and data processing. Both are skills I have in abundance.
The patients that need the most service are cut out of the private payer system and rely on state and national payers. The CMS system is the federal data system and is a much easier nut to crack than than the myriad of state medicaid systems that address the poorest citizens.
Let's start with Doctors and Clinicians - National Downloadable File
Quick and Dirty Data Exploration with Tableau
How can we use data like this to improve lives?
Target new providers to need by searching for new master-level clinicians that will be more likely to have slots
How would that help? Appointments can be 6 months out for new patients. Payer inefficiency boxes people out of appointments that they need. They end up in emergency rooms and urgent care centers and are often unable to pay. Thereby putting the burden on hospitals. Better targeting of new providers with lower-credentials like the master's degree can get patients to care providers sooner.
What would need to be done? The public database needs cleaning - zip code looks like this: 824843438 with the zip extension included. Many of the credentials are null.
The clinical social worker is the powerhouse of the mental health industry. Most of the clinical details here are null. Good data management would have better credentials.
Together, zip and credential can be used to find health care deserts: Places where there is a huge need, but little supply.
Building an Interchange Data Dictionary
Data is only as good as the infrastructure behind it. One of the best first steps to building data infrastructure is codifying the Data Dictionary. PDF dictionaries are unfortunate, because they block automation. JSON is a far better format for data dictionaries.
I have a script that parses the Provider Data Dictionary and produces a JSON file.
It generates a readable Data Dictionary in JSON that can then be used to build out a DDL in your favorite database.
Here we see both the coded values and the English descriptions. Building analytics is akin to writing and good analytics systems boil down to asking questions in natural language of data.