Thursday, January 16, 2014

The Failure Of Google Health Seems To Have A Lot Of Commonality With The Present PCEHR Situation.

This appeared a little while ago.

Why "Google Health" failed.

In his Second Opinion column, Dr Constantine Constantinides from healthCare cybernetics looks at why “Google Health” failed…the Electronic Health Record “Death Trap” and its implications for Medical Tourism
If you were wondering what ever happened to the Google Health initiative … Let me provide some background to the death of Google Health.
What Google announced was this:
"Google Health has been permanently discontinued. All data remaining in Google Health user accounts as of January 2, 2013 has been systematically destroyed, and Google is no longer able to recover any Google Health data for any user. To learn more about this announcement, see our blog post, or answers to frequently-asked questions below"
Even Google couldn’t master the concept of the Universal eHealth Record!

There can only be one Electronic Health Record (EHR)

Meaning… the “single-instance storage” concept and practice.
Late in life, I came to the conclusion that in many instances, success hinges on the “single-instance storage” concept and practice. With healthCare cybernetics (hCc), I am still struggling with the challenge of adopting and implementing the principle and practice.
For the EHR to succeed it must, likewise, be based on the “single-instance storage” concept and practice.
Practically all who claim to be in the business of “Medical Tourism” seem to know about the Electronic Health Record – eHealth Record – EHR - (and some feel confident enough to express informed opinions on the subject). Some who are in the medical tourism business feel even more confident enough to want to establish a proprietary eHealth Record System – to be adopted and used “by one and all”.
Everyone wants to be the keeper of your eHealth record!

What are the challenges?

I emphasise …single and universal
The Universal eHealth Record continues to confound even the best brains in the field of health informatics. And here, we are talking about the Universal eHealth Record – and not an eHealth Record system designed to satisfy the limited (and short-sighted) needs of a hospital or medical practice.

Why Google health failed

For an eHealth Record System to be “universally” and practically useful, it needs to be: 
  • Language-neutral.
  • Comprehensive (complete).
  • Stored in only one place, and updated at that place (the “single-instance storage” concept and practice).
  • Universally compatible and interoperable.
This ideal eHealth Record System continues to elude us.
No single eHealth Record system yet introduced has been adopted by the number of users required for the “tipping point” to be reached.
And of course, the introduction of more and more eHealth Record systems, specifically for medical tourists, is not helping matters or the health consumer.
The full article (with lots more on medical tourism etc.)  is here:
While not sure I am totally comfortable with what is said here there is certainly a sensible basis  here and the article is pointing out what has always been important and which the PCEHR just ignored.
I wonder what happens next.


Enrico Coiera said...

Let me beat that old drum from Chapter 1 of my textbook:
All models are wrong.

This means there is and can never be:
a universal health record
a universal standard
a universal knowledge base

or put simply - not built technology is universally useful for all tasks.

Pick your population, your task, and do as well as you can for now, knowing a new solution is already out there somewhere to make your current one redundant. And prepare for that moment of redundancy ....

Informatics 101.

(Sorry must be the heat, am getting cranky and pedantic)

Anonymous said...

In the abstract Enrico is right, but ... we can do sooooo much better than the attempts so far.

BTW Google Health failed because their health experts thought it was just another XML data problem. They didn't understand the health domain at all. No amount of money can fake understanding health informatics (unless you hire some proper health informatics experts).

Anonymous said...

What Enrico says is true in the abstract. But we can do sooooo much better than the current attempts (by major corporations) at shared EHR. And the technology is there.

It's not about finding some silver bullet model / architecture, it's about finding one that solves the most requirements, in a flexible way.

BTW Googleheath failed because Google didn't understand the health domain at all, they thought it was just an XML data problem.

Anonymous said...

As heretical as most cranky, pedantic health informaticists will find it, there's something undeniably charming about the idea explored in this article:

Enrico Coiera said...

Its charming now, but wait until the committees get to it.

We are going to need:

1 - a core minimum data set agreed for the "about" box. I'm sure we could limit the set to several hundred data items, but we might need a few years to agree on which ones. We'll get back to you as we progress ..

2 - Ensure compliance with national privacy laws for the privacy settings bit - we might also need a stakeholder consolation, and then some consultants to evaluate on the consultation - and perhaps also need to revisit the opt in/out debate, before we can move ahead with this.

3 - I think this whole thing about inviting other clinicians and sharing with them is just as fraught, so repeat the whole enterprise described above for that. Maybe use a different international consultancy house to do this?

4 - The status updates are just alerts and we know how annoying these are to clinicians - but we also know how dangerous it is to turn off some alerts. Hmm ... so repeat step 2 above to work out how alert settings can be safely customised, and we probably also need some standard for alert terminology and structure - signed off by all parties ... yay another committee

5 - repeat 2 for tagging behaviours

6 - repeat 2 for notifications of new data arrivals

7 - I forgot terminology! Sorry. We must standardise that. You may have to wait until we do but we are oh so close now. I can feel it in my semantically interoperable bones.

Good, I can feel progress already. Facebook UIs are the answer, but they do need a good dose of bureaucratised reality before they are ready for prime time. This time we really will get it right, right? Onwards ...

Oh, by the way, where will all that data be stored? Facebook operates its own database, but clinical data is distributed .... let me see if there is a committee for that ....

Bernard Robertson-Dunn said...

IMHO, health is not about data, it's about health outcomes.

Outcomes are achieved by implementing processes - both human and automated.

Processes are supported by access to data.

So who is defining health processes - both existing and potential?

It seems to me that focusing on health data is putting the cart before the horse and could be one of the major reasons why so much health IT fails miserably.

Sort out the processes first and then identify what information you need.

If you can't standardise processes, then you certainly can't standardise information and data requirements.