Mayo Clinic Visit with Robert X. Cringley

[rofthorax]

Read my comments backwards as it seems this shows newest messages first..

[rofthorax]

h t t p : / / t i n y u r l . c o m / 5mabcu

Remove the spaces in that line and put it in your address bar.. This link will take you to the OMG PIDS specification.

h t t p : / / w w w . o p e n e m e d . o r g /

Remove the spaces from this url above, this is a reference to the Open source implementation of the PIDS specification.

You will need experience with Java to utilize and develop it. Contact me through my youtube account for more information.

[rofthorax]

On the OpenEMed website I just noticed they didn't use the PIDS acronym, it is "Person IDentification Service" written out. There is another specification COAS (Clinical Observation Access Service), and RAD (Resource Access Decision), and Terminology Query Service (TQS).

COAS specifies a way to store and access patient records (yes really!).

RAD is security.

TQS correlates healtcare "codes" so records can be transmitted oversees to say french hospitals, say you get in a car wreck in paris.

[rofthorax]

So for more information, go to the OMG's website, and look up the PIDS API specification, there should be plenty of adobe pdf's on the PIDS API.

There is an open source implementation of PIDS, which is not perfect but could be worked on and developed into something useful, on source forge, called OpenEMed. It was developed by engineers at the Los Alamos National Laboratory as part of a project called "Telemed" (rumor: was pulled as soon as Microsoft vendors felt it would compete MS-office).

[rofthorax]

Note, PIDS don't autmatically correlate ID's, what they do is define a distributed networkable API for integration between PIDS implementations. Each implementation will have its own way of correlating information. Like the use of fuzzy recognition like "this looks like the person, but it's but only 60% is matching, I'll leave a note to the administrator to determine if it is a match".

So the solution doesn't have to work unguided.

[rofthorax]

CORBA is not just a abstract data format as that's all XML is, CORBA is a distributed architecture for object interfacing. This means that say I have a clinic with two PIDS correlators on it, all I need to do to add a hospital is to install PIDS there, and connect it to the clinics, the PIDS systems at the clinics will find the PIDS at the hospital and begin correlating records using patient biometrics, WITHOUT a master patient index (meta-database). And all without disobeying HIPPAA.

[rofthorax]

I meant to say two clinics with each one having a PIDS correlator. Whoops.. Actually the technology scales because it works in a distributed fashion (no central system is needed to permit connectivity between the sites), like how bit-torrent or napster does but a bit more intimate with a lot more accountability.

[rofthorax]

BTW, PIDS is being used to correlate mexican immigrants with criminal records now. But I doubt it is being used to a great degree in the medical institutions.

Note, CORBA, the integration technology behind PIDS, like XML allows for information to be shared between unrelated technologies, however CORBA is much more efficient. The GUI, Gnome, for Linux and Netscape 4+ has a CORBA backend. CORBA was designed originally to connect processes in a distributed network (for parallel computing).

[rofthorax]

BTW, if you are a mexican immigrant and wish to avoid being recognized, best to get rid of any identifying qualities, like tattoos, missing fingers,name, eye color,finger prints, blood type... Good luck..

[rofthorax]

See part of the problem is the databases at each hospital and clinic have their own record id's.. The only way to correlate records outside of PIDS, is to use another database to relate record id's at every hospital. But this only gets more political and cumbersome as it scales, thus it doesn't really scale. Biometrics (eye color, blood type, fingerprints, etc) plus address and identity information is the only way to make it work in a distributed fashion.

[rofthorax]

See, what Medical Manager does is it prevents hospitals from sharing a record type called a ID9 (XML based format that gives non biometric patient identity, like name and address), we were charged 5000 dollars to access their software's ID9's so we could integrate PIDS in to allow patient records to be correlated. 5000 dollars for address information. PIDS uses any information available plus a correlating algorithm that is implementation independent, so biometrics or address, each contributes.

[rofthorax]

Look into OpenEMed at source forge, it is a open source PIDS implementation. Talk to David Kilman he pushed the development of this electronic patient record system, I worked with him. PIDS is a biometric database independent CORBA implementation, it permits patients from multiple clinics to be correlated by biometrics so that records from multiple institutions can be recognized and related. The OMG has plenty of related API for this.. "Medical Manager" is the microsoft here.