Security, stability, and scalability; these are the big three of handling healthcare data. Our technical team took a hard look at the Tolven data model with regard to the big three, and found that the architects had surpassed our expectations in every area. Here’s what we found.
The database was designed with encryption of health data in mind – rather than having been implemented after-the-fact. This means that it is a more efficient implementation, because the system implementor can configure what gets encrypted and what does not. Shouldn’t everything be encrypted? According to HIPAA, you must protect Personal Health Information (PHI), which is a combination of data that can identify a person as an individual and information about that person’s health. The reason this is important becomes evident when you consider that encryption is good for protecting data, but it is terrible when it comes to accessing data; it is slow. Therein we find the gleaming difference of the data model! All PHI is stored with a multi-layered, robust encryption to ensure that it is safe, but then you can choose to extract key data elements and store them unencrypted as indexes to make accessing the information as fast as possible.
We love message queues! The data model is driven by transactional message queues (implemented in JMS), which makes for a very solid data-handling foundation. This is layered on top of the PostgreSQL database, for the open source stack, or Oracle for those who prefer it. Both of these RDMS powerhouses have a strong track record and vital community of users.
Normally, when I think of scaling a database, I think about the advanced features of PostgreSQL that support managing multiple instances and replication. This type of scale is certainly valid, and essential in the realm of healthcare data. Within 2 years of the launch of our EHR system, the database had grown to nearly 100 GB and was only growing more rapidly each time we deployed a new feature and brought more clinical users on board. There is, however, another type of scaling that might be better described as diversity of data. This type of scale takes place as the clinical processes with in an organization are refined, new technologies are introduced, and human beings find new ways to provide and quantify healthcare. Say, for example, that your nursing staff are collecting new data in their daily rounds. In the golden age of relational database design, the solution to this need was to modify the database and add columns or related table to accommodate new data. Enter the document-based data storage in the Tolven data model. The document is an XML format (HL7v3 RIM-based) that is processed by a rules engine (JBoss Drools) in order to create indexes and metadata. Do I want to collect this new observation made by my nursing staff, just add it to the document. Do I want to index this data element and drive workflow based on its value? Just add a rule or two to the rule package. No database modification required. This is the scaling that makes a system that will grow with your organization, not only in size, but also in refining the data that drives healthcare.