Episode 15: One Version of the Truth

One of the problems about data it that much of it is, basically, crap. Go onto Experian and check your credit sometime. Go back through any set of records and the number of errors you will find will astound you. 

If you want to figure out which company is selling your “personal information” my favorite trick is to put a different middle initial (or some small change) on the forms. Most systems can’t correlate well enough to fix it so you can get insight into how prolific the information business really is; it is also a great mail sorter.

Whole companies have evolved to do “data cleansing.” Their goal is to try to find the correct information amongst a mass of data (like isolating your correct address) so they can send you and me another credit card offer – oh joy. What I would really like is a “business intelligence” application that would determine that I haven’t applied for a new credit card in 15 years and stop sending them! 

Data cleansing has gotten pretty good, very good in fact. There are some areas, however, where this approach will simply not work. 

Healthcare is one.

I gave a speech today on building Secure Information-Centric IT systems for Healthcare. This is an area I am passionate about. I am passionate here because there is so much that we could do to improve our healthcare systems if we could find a way to better leverage technology. 

You probably wouldn’t know it but the fifth largest cause of death in the U is “avoidable medical errors” --- 98,000 people/year! That’s higher than deaths from motor vehicle accidents, breast cancer, or even AIDS.

Why? 

It’s not because we don’t have the data. I believe it’s actually because we have too much data…and not enough information. We lack “one version of the truth” or what I call “Authoritative Data Sets.” We have the data locked in “application silos.”

My son has gone through a number of surgeries. I am always totally shocked, but now expect, to be asked for his basic medial information (like allergies) even minutes before his surgery. I think – what if I forget something? Is my son’s life dependant on my recalling the right six mediations he is allergic to? There has got to be a better way. 

For critical data – we need to have one version of the truth. Essentially, this means that there needs to be fewer sources for given information. Data should not be “owned” or localized to single applications; it must be delivered to any and all authorized individuals/applications from an authoritative data record.

Think about what happens when you go to a new doctor or dentist. Usually they won’t even ask for your old records, and even if they do, they will still hand you some forms the size of a small novel to “fill out from memory.” Have you ever had surgery? If so, when (mo/day/yr)? Was it raining that day? And on, and on… If you are like me, you just guess on the forms and you thereby create another independent, inaccurate record that will not be properly managed or maintained. 

What I want to know is, since all of that data exists, why can’t I simply “authorize” my new doctor to have it?

The concept of an authoritative data model is simple. If a person needs certain information about me, I would no longer fill in a form – thereby creating a new record. Instead, I allow a person access to an “Authoritative Record” to retrieve the information. 

The trick is that, as the information itself changes, only a single record would be changed. Applications needing that information would be required to pull from the Authoritative Record rather than creating a separate record.

Let’s take a very simple use case. What if, instead of telling all of my utilities, banks, friends what my address is, I tell them where they can find my latest address. If you have ever moved – could you imagine how much time would have been saved if you went to a web site and changed one form to move? 

Clearly, with any use of information, we need security and privacy, but we have the technology available to meet these needs. Getting the right information to the people who need it is what we need. This is going to be a huge challenge and costs are significant - but the payout is enormous! How much should we invest to fix something that could radically reduce the 5^th largest cost of death in the US?

If you would like to look at this more - I posted the slides from the speech (sorry no one recorded this one) on the presentation links. 

You will definitely hear more from me on this topic.

Mark…