Family Health Guy

In which Sean talks about HealthVault and other cool ideas in Personal Health

CHDI and Amalga --- two great tastes that taste great together!

CHDI and Amalga --- two great tastes that taste great together!

  • Comments 0

One last post about Health 2.0 this year. In addition to all the Blue Button cheerleading, I had an opportunity to showcase Microsoft Amalga combining (anonymized) clinical information with CHDI indicators to deliver some pretty cool insights. For my money, this is the where CHDI has the potential to really make a difference --- helping providers deliver better care.

We call Amalga a real time intelligence engine --- it allows an institution to assemble a "data asset" consisting of all kinds of relevant information. Clinical, operational, financial, environmental, whatever --- Amalga sucks it all in and then provides tools to synthesize the data into useful views that join across domains, in real time. Just for example, you might construct a view that combines financial and clinical data to predict which patients are at risk for bounceback and would benefit from additional attention before being discharged.

This is what we did with CHDI --- combine key indicators with specific clinical data to look for correlations between where people live and their hospital experience. The coolest thing about CHDI is that it's reasonably granular, with many indicators down at the county level, making the join meaningful.

The first step was to make it really easy for Amalga customers to integrate the CHDI data sets into their local databases. We wrote a process that "cleaned up" the CHDI data and then pushed it up to a special queue in our cloud based storage system Windows Azure --- then created a plug-in package for Amalga that automatically pulls the information down and loads it into the local store. The end result is that with just a couple of clicks, any Amalga customer can now have access to the CHDI data. Pretty cool!

OK, so here's the fun part. We have a really great set of demo clinical data that has its origins in real hospital data, but has been fully anonymized so that we can use it for testing and demonstrations. I loaded up an Amalga instance with this data and the CHDI stuff and let one of our internal docs loose on it. The results were pretty fascinating!

As a thought exercise, the question we asked ourselves was this: Does living in an area with poor air quality increase the amount of time that patients stay in the hospital? To dig into this, we created a view that joined hospital visits, including a column for length of stay, with the CHDI air quality indicators:

From here we can select a cohort of patients and use Amalga's "quick stats" function to compare average length of stay for the whole cohort versus the length for those patients that live in an area with unhealthy levels of particulate matter in the air. Not surprisingly, we do NOT see any real difference here.

However --- let's look a little deeper. What happens if we filter down the list of visits to just those with complaints related to asthma? The same quick stats on this cohort are much more interesting! Asthma patients from high particulate areas seem to spend a full half day longer in the hospital --- almost a 10% difference! Now, there are a bunch of reasons this could be the case --- but it's sure an interesting observation and a great example of the kind of insight that's hiding away in our data just waiting to be discovered.

Now, this gets really exciting when you take the next step and apply these observations to real clinical care. For example, you may triage asthmatics from risky areas higher than other patients. Or if you have a new mom give birth and she lives in an area with poor nutritional stats (high obesity, limited access to fresh foods, etc.) --- you might send her home with additional educational materials about healthy eating for her new family. As we move towards a world of payment reform and ACOs, this kind of population management can make a real difference.

Later this year, HHS is going to release a ton of additional data as part of CHDI ... including some really granular stuff from CMS. Amalga is just perfectly situated to turn this data into real clinical value --- two great tastes that taste great together indeed. I am loving the momentum --- wooo hooo!

Leave a Comment
  • Please add 1 and 5 and type the answer here:
  • Post