Electronic Health Records and the Challenge of Unstructured Data

By Sheldon Needle

The real problem of an established medical practice moving into the realm of EHR is not the cost of the medical software package; it is not the training necessary for staff; and it is not security and backups.

The real problem of moving into EMR/EHR is the problem of unstructured medical data.

If you are involved in a new or relatively new practice, this is a no-brainer. Begin with a serious search to compare medical software vendors who are available to answer your questions honestly. It is not truly so difficult to get on a friendly medical screen to enter your patient’s blood pressure or lab test values. You can get used to that.

Neither is it difficult to take notes on a notebook that upload to the EHR system.

The real problem is taking your notes and dictation on a patient that go back 15 years and finding a way to get his possible symptoms, his worry about IBS, his headache history, and his worries over his children into a metrically available rendition that that does not take you or a member of your practices days to decipher. These notes are usually on dictation, hand written notes, and referral letters.

The concerns are many: this can take what feels to be forever, and the anxiety issues and unclear symptoms may not translate easily into metrics but may be critically important in future diagnoses.

There are two critical questions here:

1. 1) Is it worth it? and
   2) If it is worth it, what to do to make this work efficiently?

In the long run, it doesn’t even matter if it is worth it. It will happen. Medicine as well as the rest of our cultural world, is becoming electronically-based whether we like it or not. But in the long run, it is worth it. Think of a patient going in to the hospital after a car accident, all by himself, and having all his data available to the admitting doctor in an instant: blood type, history, etc.

Think of a patient being referred to you, the specialist, and having all his patient history available in less than a minute. What a time saver! What insight!

Medical informatics has a number of methodologies it is using to translate unstructured data into useful and structured data.

Three basic methodologies exist to accomplish this:

• String matching
• Natural language processing for Medicine (NLP), which uses syntactic rules in extracting data from text documents
• Concept-based indexing which uses data base codes to group and relate medical concepts

These methods will be refined, utilized, and integrated in some way into most decent medical vendor software packages over the next few years. For you the physician or practice manager, this may start to pay off in a while, but you still have to get from hand written records into the database.

The obvious way to proceed makes use of our culture idea of, “going forward”:

1. Start with today’s records being input into the database electronically – this is the easy part.
2. Then get help in moving 1 year of back data scanned and automated. Get someone technically savvy and talk to the support people whose EHR software you are considering about OCR (optical character recognition) software that may be available from vendors.
3. Most vendors of decent repute will have voice recognition software incorporated into their total EHR solutions. Have them demonstrate how well it works in moving data into their files.

The real message to practitioners moving to electronic health records is, don’t look at the top of the mountain when you start climbing, just put one foot in front of the other. Delaying the climb will not get you anywhere, but starting the march will move faster than you think!

Accounting Software | Construction Software | CRM Software| Medical Software | Manufacturing Software
Testimonials | FAQ | Tell A Friend | Industry News | Find a Consultant | Privacy Policy | Links | Site Map