Though electronic health records (EHRs) have been around for more than 30 years, it wasn’t until 2010 that healthcare systems and individual practices began looking at them in earnest. The idea behind EHRs is to provide universal access to records among doctors, health clinics, hospitals, etc. in order to make it possible to treat patients regardless of where they seek care.
To date, EHR software packages have been anything but universal in their application. As such, compatibility problems remain. And where there are compatibility issues, there is also an inability to take advantage of the big data strategies that were supposed to follow on the heels of EHRs.
Thankfully, there are two technologies capable of improving EHRs dramatically. Those two technologies are natural language processing and medical record searches.
Different Kinds of Data
When you step back and analyze why healthcare seems to be struggling with big data and EHRs, it becomes apparent that there are two kinds of data that have to be accounted for. The first is structured data. This is data entered into record by way of things like drop-down menus and tick boxes. This sort of data is easy to standardize, collect, and analyze thanks to its structure.
The second kind of data is not so easy to work with. It is known as a free text data. Think of it as hand-written or electronic notes jotted down by a doctor during a patient visit. Free text documents generate a ton of data that could be helpful to everything from diagnostics to understanding health histories. But how do you access that information from an HR system? And once you have the information, how do you analyze it?
Natural Language Processing
One of the things software developers are turning to now is natural language processing, a specific discipline within the arena of computer science tasked with developing ways to take advantage of human language as naturally spoken or written. It is by no means an exact science.
A natural language processing (NLP) system should ideally be able to scan a portion of written notes and glean valuable data from it. The system would have to be able to recognize certain words, sentences, phrases, etc. It would have to be able to analyze sentence syntax and structure.
Unfortunately, NLP requires a tremendous amount of manual input from grammarians just to build a functional system. But even the best functioning NLP systems are far from perfect. The good news is that developers are getting ever closer to creating a reliable NLP system for health records.
Medical Record Searches
A more promising area, according to California-based Rock West Solutions, is using Big Data sets to assist with medical records searching. It is a technology that has been around for more than a decade. Medical records searching is based on the principle of finding desired data and then allowing the searcher to decide what to do with it.
A system known as EMERSE, developed by the University of Michigan, does just that. It is a system that has been programmed to recognize specific medical terminology in free text documents. Users can search specific words or phrases just as they would with an internet search engine, then use the results of the search as they see fit.
The combination of NLP and medical records searching promises to do a lot for EHRs by making better use of free text data. The only question is how long it will take. The sooner EHR systems make maximum use of the technologies, the better they will be at doing what they are designed to do.