MPI's are not limited to hierarchical architectures, they can also be federated. What you describe is detailed I think rather well in the CORBAmed PIDS specification. In fact, one does not even have to have an MPI but could use the matching algorithms on each transaction. What does have to be "mapped" are the trait concepts (e.g. DoB to DOB, name to name, etc) and establish how close the trait values have to be to match and give them weights to contribute to the overall score (high cumulative score means a good match, low score means they're definitely different, an intermediate score means ... get someone on the phone to check further). For example, names can have soundex as well as synonym lists (Bob matches with Robert, etc.)
-----Original Message----- Greg Woodhouse wrote: That's not an insignificant problem because different organizations are bound to use different identifiers. A Master Patient Index (MPI) is a good solution when everyone uses the same index, but what if they don't (or can't)? HL7 takes an interesting approach, allowing registry identifiers to be part of the unique key. But how are registries identified? Is there a body that assigns unique identifiers to VHA, Kaiser, CHCW, and so forth? The problem is not at all unlike the use of protocol numbers in TCP/IP (assigned centrally by the IANA) or domain names (which are not centrally assigned or managed, but instead are handled through a hierarchical series of registries). Ultimately, I believe we will need some type of hierarchical designator and a set of open standards describing how they should be managed. --- Cameron Schlehuber <[EMAIL PROTECTED]> wrote: > The key is to have a "unique key" that both systems use to identify > each > record uniquely. The problem of duplicates arises when ambiguous > identifying traits are all that are used. Names, date of birth, etc, > can > all be entered with slight variations for the same person and can > also be > found to be the same for different persons. That's why once a unique > ID is > assigned to a set of traits for a given person, using that ID (or > "key" in > table records) for solid identification becomes very important. When > you > don't have such a "key", the process of matching up the data has to > be done > all over each time. ------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://sel.as-us.falkag.net/sel?cmd=lnk&kid3432&bid#0486&dat1642 _______________________________________________ Hardhats-members mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/hardhats-members
