Howard McCollister cut
I see where you are coming from now. You primarily work with large systems. Don't think you deal with the majority of patient populations though. Primary health care (suburban doctors) deal with the majority of patients. Mom and Pop operations as you call them. On a typical day in your country many millions of people seek primary health care advice while far fewer numbers need hospital care. Helping GPs isn't a big task, or a particularly glamorous task, but a cheap effective practice management system helps more people than a complex hospital system.
Addressing your comments about complex systems: One solution that was effective in your country was to leave the legacy computerised systems in place and tie them together with glueware.
The middleware used in that instance was a broker computer running apache. Legacy systems had modules added which could respond to http requests. The result was the unification of hardware incompatible systems into a system which could produce any combination of information in the hospital.
A doctor could call up a screen which displayed patient monitoring information, from intensive care, patient history from records, health insurance information from finance and the doctor's personal notes. Better decisions on patient care could be made taking into account all relevant information.
Of course here in Oz we have universal health insurance and much of the information needed for care choices in the US is not critical for us.
But this example shows that it is possible to develop a highly integrated system, without replacing legacy equipment, at very little cost.
BTW, don't knock interpreted languages such as PERL or generic software like EXEL. Perl can handle the procedural aspects of a system very efficiently, leveraging the power of the DBMS when needed. It can link together hardware and software that were never meant to interact by their designers, and do it well.
Exel can be manipulated by hand, which is how it is used by most people. It can also be used as a software service by other programs. I have a couple of neural nets inside excel workbooks on my desktop right now. I'm not smart enough to write a neural net but I can bolt a prewritten one onto a system with glueware.
Good solutions don't have to be complex or expensive. They just have to meet the clients needs.
DM personal opinion only
