Wednesday, September 25, 2013

Applications of BigData in Health Care Industry

In my previous post, I discussed about Big Data and how it is being used in today’s world. In this post, I will discuss more on how Big Data is being used in a particular industry such as healthcare. These days terms such as NoSQL, MongoDB, CouchDB are being very frequently used. Have many of us wondered why they are being used or what are its benefits? As discussed in my last post, data from external sources such as social media data is becoming prominent rather than the internal data which companies used to rely on 10 years back. It is a challenging task to restructure the database schema and data warehouse to fit the external data. It is not a feasible option as well since external data comes from various sources and they do not have any specific format. How can this data be tracked and stored to analyze results? To store these kind of data, we require a non-relational database.

The terms MongoDB, CouchDb are types of non-relational database which does not have the relational structure. Healthcare is a major industry which uses BigData and the necessary applications to track and analyze patient records. While many companies use Data warehouse and star schemas to perform prediction and reporting, there are few who use the traditional NoSql databases. One example to illustrate this point is the use of MUMPS database in IT Healthcare companies such as Epic Systems. 

MUMPS database is a traditional database which was used back in 1950. Its prevalence was lost because of the invention of SQL and RDBMS. But recently since external data is increasing these traditional sources have gained their importance. It is a hierarchical database unlike the relational database. It is very useful in the case of healthcare industry since it helps them to maintain the data efficiently without placing a constraint that the database should be in 3NF. To explain this point more clearly, if the hospital wants to track the behavior of patients along with their frequency and cause of visits. Using the relational schema, we could have a relationship between patient and visit. Each visit is related to a procedure. It is possible that a patient can have more than one visits and each visit can have multiple procedures. Maintaing these in a relational database will require the tables to be in the 3rd Normal form. Because of this constraint, it becomes difficult to maintain multiple details such as phone numbers of the patients. Creating separate attributes for the phone numbers can solve the issue but it becomes a cumbersome task maintaining the NULL values for phone numbers (patients who do not have multiple phone numbers). Creating a hierarchical database can be a better option since it does not require the tables to be in 1NF.  

This post just gave a brief overview of why traditional databases such as MUMPS are being widely used in the healthcare industry and their use.  In my following posts, I will discuss more on Big Data, Business Intelligence and their applications. 


No comments:

Post a Comment