[Previous Months][Date Index][Thread Index][Join - Register][Login]
[Message Prev][Message Next][Thread Prev][Thread Next]

Re: [IPk] Programming the Pump

> Thanks for this John.
> I don't have a com station thingy.  But if it can be used to
> programme the pump, I would look at getting one!
> Michael - by 'automatically' I don't mean that the computer takes it
> into its own head to tell the pump to do something.  It's like
> automatically dialling a remote computer through a modem: I, as the
> user say 'go to the Demon / English Heritage / Freeserve /
> University of York server, and the computer takes that instruction
> and translates it into the beeps and burrs which will get me hooked
> to the computer at the other end.  It doesn't decide all by itself
> that I need to read emails at Demon, edit a database at York or
> whatever!  The computer didn't decide by itself what the telephone
> number for Demon is, either.  I told it.  
> Similarly, the pump software I'm envisioning won't decide that I
> need to exercise, get a cold, or whatever.  I will tell the computer
> that I need a certain basal pattern, and it will do the beeps and
> burrs (or rather, the infra-red flashes) which will save me a lot of
> button pushing. It won't decide by itself what the pattern for
> exercise is, either.  Like Demon's telephone number, it will know
> because I've told it.

The question is, how is the determination made about what the basal 
levels should be?? The program to do that now is a "human" that 
carefully intreprets the data, throwing out bad data points and 
things that look unreasonable -- either you or your doc or a 
combination. Simply plugging numbers into a computer by rote and 
extracting basal rates would in my estimation be very unsafe. No 
amount of safeguards will eliminate the slip of a finger on a number 
or the assumption on the part of the user that the data is good when 
in reality it is faulty. Using a computer tool as and aid is one 
thing, having it take over the job of determining safe level of 
potentially fatal medication administration is entirely another. 

The "closed loop" system scenario which you might mention next is not 
a good comparison. The computer in this case constantly monitors and 
adjusts without ever really looking at absolutes, just incremental 
changes. However in the scenario you paint, it is a set-it and 
forget-it operation done only once in a while. 

I think that the opportunity for error is very high and knowing that 
implies large potential liability when errors occurs. The human 
body + insulin + carbs, even when modeled as a relatively linear 
process is still a 2nd order system. I have worked with this kind of 
model -- see the example on the HOWTO page -- an know that with 
marginal data it gives results that can easily be out of range but 
not obvious except that the output is nonsense. Try it, you'll see 
what I mean.

All that said, I think such tools are very useful and can help a lot 
in establishing and tuning basal rates. I'm very skeptical that this 
can or should be done "automatically". Doing it that way implies that 
the answers are "right" and can be trusted. That's human nature. 
Afterall, you know how stupid people can be at times :-)


> No computer-as-god/doctor at all: all keys pressed by a human
> (regrettably - I have found that virtually every time I get a non-
> perfect bg there is human operator error - pump error is very rare,
> and I'd gladly hand over control to it!)
> Best wishes to all,
> Pat
> dm 30+, 508 1+, looking at a thesis for which I am very much afraid
> that over the next six months I must learn xtml, and establish an
> SQL database server. Wish the computer _could_ think for itself!
> In message <l03130303b9558aa57f08@[]>, John Neale
> <email @ redacted> writes
> >Pat - MiniMed is driven primarily by the US market. We are just a spin-off.
> >If you understand how the legal restrictions of the US healthcare market
> >work, I'm sure you could earn a lot of money as a lobbyist on Capitol
> >Hill...
> >
> >But I digress.
> >
> >Have you got a Comm-Station thingy? If so, you can get your computer to
> >talk to your pump. If you have the skills to program a spreadsheet to
> >produce nice curves etc, you surely have the skills to write the software
> >you want. It seems hard because you've not yet done it. Choose your
> >prgramming language, and let us know how you get on.
> >
> >I've done a bit of programming in my time, and I personally think that for
> >various reasons it is easier to program things directly into the screen on
> >a pump, than to connect the pump to a computer and do everything remotely.
> >But that's only my opinion.
> >
> >John
> >
> >--
> >mailto:email @ redacted
> >http://www.webshowcase.net/johnneale
> >----------------------------------------------------------
> >for HELP or to subscribe/unsubscribe, contact: HELP@insulin-pumpers.org
> >help SUPPORT Insulin Pumpers http://www.insulin-pumpers.org/donate.shtml
> -- 
> Pat Reynolds
> email @ redacted
>    "It might look a bit messy now, but just you come back in 500
>    years time" (T. Pratchett)
> ----------------------------------------------------------
> for HELP or to subscribe/unsubscribe, contact:
> HELP@insulin-pumpers.org help SUPPORT Insulin Pumpers
> http://www.insulin-pumpers.org/donate.shtml
for HELP or to subscribe/unsubscribe, contact: HELP@insulin-pumpers.org
help SUPPORT Insulin Pumpers http://www.insulin-pumpers.org/donate.shtml