A little while ago I wrote about using Splunk, a nice IT systems log aggregator and search engine, in the healthcare sector. I just spoke with the product manager in charge of the tool at Splunk and I liked what I heard even more: they have a solid design, good APIs, and an open approach to accepting almost any kind of ASCII data and creating a searchable index and relationships between data. I spoke to them about how they might be able to support ASC X.12, HL7, and other healthcare data formats and it seems like it’s not only possible, but fairly straightforward.
I was thinking about a system for hospitals or other large healthcare institutions that would be able to act as a data sink for all data: financial, administrative, clinical, medical device, PACS, etc. All those data streams could be tagged, tokenized, and stored in the Splunk server. Then, we’d be able to do searches like “show me all data dealing with Physician X” and the system could do a full text Google-style search on the Splunk data and pull back information in one view with time and system information about where it was created.
A master person/patient index would even be pretty easy to setup if we could do a bit of de-duplication of data on the way in. I think the biggest benefit of something like this would be to use it as a data crawler and pull together everything in one place so that we could do forensics on who saw what & when, get all related documents dealing with a legal case, or use it for aggregating patient data in one place.
The piece that I’m still thinking about is how the data storage (which is plain text) would affect privacy and security but I think that could be worked out if we could come up with some good use cases.
Anybody interested in working with me to give this a shot at their premises? Comment or contact me anonymously. Let me know if this is a crazy idea or something worth considering.