December 24, 2009
Posted by Elizabeth Han
The Search for Better Search in EMR
The success of electronic medical records (EMR) depends on much more than the EMR itself, but intuitive, smart software is still important. In this post (Merry Christmas!), I will talk about one tough issue in particular, the search problem, and why a “finder” like Google Wave (June, November) might be a good idea.
Larger electronic record projects may show us what’s needed.
Yesterday, I read an article (National Post) on MyLifeBits, a kind of Electronic “Life” Record. Since 1998, Gordon Bell, a principal investigator at Microsoft, has led a project to “go paperless”, essentially scanning, recording, logging every single thing in his life.
The motivation?:
I’m not cluttered with having to remember any of this stuff. I just have to remember that it’s in there, and that I’ve looked at it.
The result is a staggering amount of information in all kinds of formats (see pie graph at right, in GB). If extended to an entire human lifetime, the total data would be worth at least 1 Terabyte, and that’s not even including video!
The problem is that “memory” must be two-way.
Something that is stored must be able to be retrieved.
In the ideal scenario, Bell wants a single search to call up “the records of a visit to the doctor, a photo of the doctor, a copy of his prescription, and the web[age of the restaurant where he went for lunch afterwards.” But accomplishing this fluidly is a challenge with the current architecture – described as “a bunch of quasi-independent parallel databases”.
The relationships are currently still filled in and maintained by human beings, whereas Bell dreams of a Google for memories.
Google for memories is actually an apt analogy.
Search isn’t physical anymore. Folders are physical. And in my opinion, folders are clunky. It used to be that to find things, we needed hierarchies and trees. Now we have tags and lists (think Gmail, Twitter lists, Facebook lists). A single item can have multiple tags, belong to more than one list. We still apply the relationships ourselves, but it’s less “fill-in” after the initial adoption and more “curation”. Because of this, it’s dynamic. We’re not just digitizing and forgetting about it.
Dynamic is also one of the first things I think of when pondering EMR. The EMRs of the future will be handling all of the file formats in the pie graph and more, from pdf forms to video chats to electronic signatures. New patient information will be coming in all the time. In such a climate, segmentation of data into folders is not enough. Although today’s EMRs have a lot of great features, we must recognize the relationships between data in as many ways as we can to facilitate search.
I’ve written a bit on Google Wave already (June, November) as an example of a tool that, if the kinks are worked out, could be useful for EMR. The advantage of a tool like Google Wave is real-time – that is, it enhances the relationship dimension of time.
Items are displayed and stored as they occur. Many different types of media can be mashed together and still take their place in a unique timeline while remaining editable at a later time. The units of Wave – blips, wavelets, waves — can be tagged and searched with the Google Power that we all know so well.
Maybe the project won’t amount to anything in the end, but search is not something to be taken lightly. As Microsoft well knows 
Merry Christmas!
-Effy
Related posts:
