On Inserting Data Blindly into Mathematical Calculators/Computers

[studiot]

Mathematics without understanding can be lethal.

 

Yet we are constantly exhorting others to use mathematics and pushing the doctrine that 'mathematics rules OK'.

 

Being human there are always those who will try to short cut the link between understanding and mathematics and technology, with its ever increasing computing power, is offering the illusion that this is possible by simply inputting data into preprogrammed systems that spit out answers.

 

Here is a recent example from this forum

 

http://www.scienceforums.net/topic/88108-please-help-with-calculating-limiting-reagent/#entry857546

 

This trend has long (always?) been there.

I can remember the scandal when the world famous Hospitals For Sick Children (Great Ormond Street Hospital) new block cracked and nearly collapsed.

This happened because data was input to a structural engineering program without understanding (taking a short cut) and the resulting concrete frame was not strong enough.

 

So, in my view, understanding must precede mathematics.

 

Yet we must understand both the nature of the subject and the mathematics we employ. Neither one on its own will do.

 

So how do we ensure that the seductive increasingly easy route is safe, going forward into the future?

[imatfaal]

Couldn't agree more. There are two routes forward; one dangerous and seductive, and the other seemingly retrograde and potentially bad for productivity.

 

The second is the method which I would prefer - and I bet you would too - teach those who are using the quick "short-cut" calculators what the inputs mean, what the outputs represent, and how the algorithm works such that they could work it on a piece of paper and actually understand the process at a more fundamental level. What will actually happen is that more safeguards and expert systems will be built into the "short-cut" calculators - this will have a positive effect of speeding up the majority of transactions and removing minor errors; but it will mask more complex errors and it will slowly denude the population of those who can spot those complex errors.

[ajb]

Just as a comment, I can be surprised when students insist I am wrong because their calculator gives a different answer. This can sometimes happen even if the calculators answer is clearly unphysical, say when doing some calculations in basic physics. Almost always it is just a problem of BODMAS and how calculators read things.

Edited March 13, 2015 by ajb

[StringJunky]

I see a parallel here with Tar's thread about automatic systems; how much automaticity should be allowed? When is it not appropriate?

Edited March 13, 2015 by StringJunky

[studiot]

 

ajb

Just as a comment, I can be surprised when students students insist I am wrong because their calculator gives a different answer. This can sometimes happen even if the calculators answer is clearly unphysical, say when doing some calculations in basic physics. Almost always it is just a problem of BODMAS and how calculators read things.

 

 

It goes deeper than BODMAS

 

I suggest you just point out to those students how many drug errors happen each year because some medical staff or other puts numbers into a calculator and can't realise they should disbelieve the answer.

 

Some of these drug errors result in fatalities.

 

As a matter of interest, imatfaal, how would or could your 'safeguards' work against this?

[ajb]

It goes deeper than BODMAS

I am sure it does, just in my limited experience students favour calculators over the lecturer! That is one thing, but more worrying is that these students are often engineering students! What happens when they graduate? Cracked concrete.

 

As for medical staff, I have no experience of working with them. Again, I would expect, as you suggest, that they believe the calculator without any understanding of what they have asked the calculator to do.

[imatfaal]

 

...As a matter of interest, imatfaal, how would or could your 'safeguards' work against this?

 

Not "my" safegaurds I recommended the other route

 

In your drug example - the calculator could be wired up with European Pharmocopeia and with NHS patients records DB (neither of which actually work that well online) - and instead of a simple dosage calculation it checks that the answer provided falls within safe prescribing limits for that patient of that age, weight, etc. that no drugs have been prescribed which would be potentially a clash of therapies, that no allergies/intolerances are found. It could then check this dosage against other dosages given by other doctors and their associated calculators for major discrepancies.

[studiot]

 

In your drug example - the calculator could be wired up with European Pharmocopeia and with NHS patients records DB (neither of which actually work that well online) - and instead of a simple dosage calculation it checks that the answer provided falls within safe prescribing limits for that patient of that age, weight, etc. that no drugs have been prescribed which would be potentially a clash of therapies, that no allergies/intolerances are found. It could then check this dosage against other dosages given by other doctors and their associated calculators for major discrepancies.

 

 

Unfortunately, it is not the prescriptions that were incorrect, it was the implementation.

 

Someone I know used to have to train and then sign off on the competency of everyone in Somerset, before they were allowed to administer intravenous drugs.

She had a 10 question paper they had to pass.

 

Periodically there was an argument with the power-that-be about relaxing her 100% correct requirement in the test.

 

Her answer was to say, "Here is the test. Which of these questions would you allow a candidate to get wrong? Every one of them is a resut of a death by incorrect doage administered."

 

Today we have 'Self certification and such tests have been dropped'

 

Sadly, drug errors have always happened, but they are on the increase.

Edited March 13, 2015 by studiot

[StringJunky]

What about pairing up prescribers and have a workable protocol cross-checking each others prescriptions before releasing them? One step of oversight should help reduce the inicidence of errors.

Edited March 13, 2015 by StringJunky

[studiot]

 

What about pairing up prescribers and have a workable protocol cross-checking each others prescriptions before releasing them? One step of oversight should help reduce the inicidence of errors.

 

 

The prescriber rarely administers a drug.

 

I have already noted these errors are administration errors, not a prescription ones.

[StringJunky]

 

The prescriber rarely administers a drug.

 

I have already noted these errors are administration errors, not a prescription ones.

Yes you did, sorry. Perhaps reduce the maximum size of packaged unit of lethal drugs.; don't allow an intravenous unit to be multi-dose. The maximum safe dose per injectable unit. It is more wasteful but which is more important; safety or economy?

Edited March 13, 2015 by StringJunky

[studiot]

Underdosing can be as serious as overdosing.

 

The medical profession has done much soul searching since it abandoned what, in my profession we used to call the independent check, and several studies to examine if such a check reduces errors.

 

Unfortunately the management and efficiency types have reduced manning levels so that it is not always possible to find another health professional competent to conduct such a check.

[Greg H.]

 

In your drug example - the calculator could be wired up with European Pharmocopeia and with NHS patients records DB (neither of which actually work that well online) - and instead of a simple dosage calculation it checks that the answer provided falls within safe prescribing limits for that patient of that age, weight, etc. that no drugs have been prescribed which would be potentially a clash of therapies, that no allergies/intolerances are found. It could then check this dosage against other dosages given by other doctors and their associated calculators for major discrepancies.

Funnily enough...

 

My stepfather worked at a hospital here in the US where they had an internal system like that. The doctors would enter a prescription through a terminal and the pharmacy system would validate that the order didn't violate whatever limits were set before it would actually show up as a valid pharmacy order. It checked patient meds, other prescriptions already issued, and a few other things. So there are at least some facilities that are seeing the value in double checking people's work. Sadly, this kind of system isn't a requirement, so it's only done on an ad hoc basis.