In my previous post, I briefly outlined what we do at the Health Solutions Group (HSG) at Microsoft. There, I also explained how our consumer solutions are built on the HealthVault platform.
Mayo Clinic Health Manager is an online tool that we built on the HealthVault platform that helps you organize health information for you and your family. And while it can’t replace the care you receive from a physician or other health care provider, it can provide personalized health guidance for every member of your family based on their specific life stage and health status. It’s free; so go check it out.
I contributed to the design and implementation of two main components of the application.
At the heart of Mayo Clinic Health Manager is a system that analyzes your health record data and provides medical guidelines and recommendations. These guidelines are based on clinical rules that the Mayo Clinic provides. So we needed to design a recommendation engine that can handle running regularly updated clinical rules against health record data.
Another area that I focused on was in data interoperability. One of the challenges of running clinical rules against health data is that the data can come from a variety of sources. Different sources tend to encode data in different ways. For instance, one system might represent that the user has “Diabetes Mellitus” with a SNOMED CT code whereas another system might use ICD-9. Yet other systems might just have a textual description, voice dictation, or scanned records. So there needs to be a system to normalize data before running clinical rules on them.
While the Mayo Clinic Health Manager operates on data from a variety of sources, it also acts as a source of data. The application provides a variety of editors that the user can use to enter or modify health data. So we also needed to take care in ensuring that data entered through the Mayo Clinic Health Manager could be used by other applications in the HealthVault ecosystem.
I cannot discuss the details of our system, but we have spent a lot of time on creating an accurate recommendation engine. We also took care of promoting data interoperability within the HealthVault ecosystem. There’s a lot more we have planned, and I look forward to the improvements to come.
