Dissecting the “Real Age” Phenomenon

Posted 23 August, 2007 in current trends, demos, problems, solutions

Recently a number of websites have been offering “real age” calculators which, upon asking a number of health/lifestyle questions, attempt to predict how long you will live. The difference between how long you are going to live and how long people live on average determines your “real age.” If, for example, you are a heavy smoker with a family history of heart disease, you might have been born 28 years ago, but your real age could be closer to 35. As a measure of its popularity, even Oprah and her ilk have been jumping on the real age bandwagon.

These real age calculators are not without their faults however.

1. No (or little) research is offered to substantiate their healthcare calculations
2. The numbers are frequently a little *too* clean (what are the chances that all bad things raise your real age by EXACTLY 1 year?)
3. No distinction is made between elements you can and cannot control
4. At the end of the survey, no action items are provided to allow the user to alter their Real Age. After all, unless you can glean some ACTIONABLE INTELLIGENCE from the results, these calculators are ultimately of little utility.

After seeing the calculator at http://www.poodwaddle.com/realage.htm, I spent a few hours reverse engineering it. healthtech’s real age calculator is an attempt to rectify the aforementioned deficits.

1. Based on XML: see the real age XML now: download and modify the XML as new scientific studies are released. add your own questions, etc.
2. Open Source: download the Real Age code and run it yourself
3. Better health summary at the end (action checklist)
4. Items are distinguished as controllable or not

RemedyMD’s tagline is “Better Data, Better Decisions, Better Outcomes,” and you might be tempted to think that better data leads automatically to better decisions, but that is not always the case. More often, it is the application of intelligent analytic algorithms (predictive informatics, if you will) which transforms the raw data into actionable information. A lot of EHR systems collect medical history, for example, but how many of them process that information to produce actionable knowledge?

What is your Real Age?!?

Health IT Blogs All Mapped Out

Posted 7 August, 2007 in blogging

What does the healthcare blogosphere look like? That is, how are all the healthcare IT blogs interconnected? To begin to answer this question, we’ll need some idea of the composition of health IT blogs. Fortunately we have HITSphere, a growing list of about healthcare IT blogs (now numbering about 50.)

I wrote a quick program that extracted the blogs listed there and for each of them, I spidered the first page for links and checked to see if those links linked back to the original 50. Then my program created a network graph that displayed the connections while scaling the size of each site’s node proportional to the number of other sites linking to it. I used PHP to do the spidering and JavaScript to graph it out. The result of all this effort can be seen in the graphic to the above right.

Did I lose anyone yet?

All this effort to grade sites based on inbound links forms the basis for Google’s PageRank^tm algorithm. However, as mentioned before, popularity does not necessarily equal truthfulness.

So is there a better way to evaluate healthcare IT blogs? How can we sift through all of them to discover those worth reading, that is, the best health IT blogs?

Well, for that we’ll need to establish some rating criteria. That criteria might include

1. Post weekly?
2. Been around for a year?
3. Allow comments? (a blog without the ability to comment isn’t a blog!)
4. Interesting?
5. Company or individual?
6. Moderation ranking system, where bloggers from highly rated blogs have greater weight

What criteria would you add (or subtract) from this list?

Be sure to click on the above right graphic for a larger picture of the Health IT Blogosphere!

UPDATE: As it turns out, this algorithm is called TrustRank. The next step is to brush off my old linear algebra skills and calculate the intra-PageRank of these TrustRank select seed pages. Eigenvectors, here we come!

Back to School (health records for kids)

Posted 1 August, 2007 in education, future tech, solutions

Summer is just about over and that means a return to school for many children around the nation. Mixed into the stack of papers your child will bring home might possibly be an immunization form from the school health office. Do you remember where you last placed the immunization cards?!? Is it time for a DTP shot? an HBV? a Hep A? If you’ve switched primary physicians, or worse still, states, your current doctor might not have a backup record of the vaccinations. In fact, “One in five U.S. children receives at least one unnecessary dose of vaccine by the age of two, wasting $15 million in vaccine cost each year.”^*

To help with this problem, many states have implemented their own electronic immunization repositories (no national database exists.) For example, in Utah, for the past five years or so, every newborn child has been automatically enrolled into USIIS (“Utah Statewide Immunization Information System”, pronounced “you sis.”) Qualified health care providers and educational organizations alike are granted access to the system. Now, after a visit to the doctor, a child’s immunization record is updated in the state database (running Oracle, by the way), which can then be accessed in report form by the school nurse.

Though I applaud the Utah Department of Health for their efforts to gather and store medical information electronically, the USIIS system falls short in a couple of aspects. For one, there is no facility for direct parent access to the system (no “parent portal.”) As well, having a separate database for immunization data apart from the rest of a persons’ medical information seems divergent from the goal of a universal personal health record. Finally, there is no access/permissions model controllable by the parent. Once data is entered into the system, it is viewable and editable by *all* of the users on the system.

On the up side, USIIS does have a web interface and an HL7 interface and USIIS can be configured to work with EHR systems. As well, it has the ability to connect to other states’ immunization record systems (recently it was coupled with Louisiana’s when Hurricane Katrina refugees came here.)

Does your EHR system have the ability to interface to your state’s immunization system?

* Source: National Vaccine Advisory Committee (NVAC)Report, Centers for Disease Control and Prevention (CDC).