Posts by Medical Cynic

I thought this would be about AI in medicine! The 'bot doesn't understand when the patient complains of said affliction!

Pension debacle

If you think the forwarding issue is bad, you should look at how badly Crapita has f*cked up the GPs' pension scheme.

An apostrophic error!

Astronauts' - more than one affected by these changes.

What a gas!

Perhaps they could work the same trick on UK Smart Meters?

This would put an end to the NHS's flagrant misuse of patient data without consent.

NHS IT and robots

The NHS has wasted spectacular sums of money on clinical records systems, with pretty poor results.

GOK what would be the outcome of any foray into AI. Robot surgery? Fine if very routine - but there's a lot of variance in the basic human design which can trap the unwary.

There's a saying in the surgical world that There's no such thing as minor surgery - just minor surgeons

As to AI in diagnostics, they may just match the standard of NHS 111 [ie piss-poor], but the complexity of diagnosis in unsorted fresh cases, with the added complexity of languages and dialects is huge. Not to mention the difficulties of clinical examination.

Risk of death

"It is possible that the transparent section is a safety feature so that its operation can be visually confirmed."

Yes indeed. You need to be able to see an air-fluid level to confirm there is fluid going into the veins, rather than air.

Risk of fatal air embolism otherwise.

This can have disastrous consequences if the lack of attention causes a crash.

http://metro.co.uk/2009/05/05/woman-bites-lovers-penis-off-in-car-crash-81436/

Presumably risk of choking for the 'other half', too.

But where's the Sideways RAM?

Where is the responsibility here?

Can the patient whose MM was flagged as 'doesn't need further Ix sue the programmers?

If so, the programmers of the app, or the programmers of the AI software that developed it?

Or just Google as an entity?

Press release reliability

It seems to me that, reading between the lines of a press release no doubt written by someone without any knowledge, that they are talking about bar-coding the packaging rather than the implant itself. I can't imagine a barcode on a small orthopaedic screw being a)possible or b) readable without major surgery. If the package is barcoded, then all the bits used in a particular procedure in a particular patient can be reliably recorded in a central database for such implants.

Then all you need to know is the patient ID and the procedure concerned.

Re MRI, inductance can cause problems in addition to heating. Mainly neurostimulation. At high field strengths, this can be a direct effect on the nerves themselves, causing tingling sensations. At all filed strengths, lengths of wire close to nervous tissue will have much stronger effects.

An example would be retained pacemaker wires. Sometimes when removing a pacemaker [eg temporary pacing no longer required] the wire or bits of wire break off and are left in situ. These are unlikely to cause problems in normal life, but in an MRI scanner, current will be induced, this in a wire that is deliberately positioned to link to the heart's conductive sysyet. The result is the scanner pacing the heart at the repetition rate of whatever sequence is running - which is generally far in excess of its maker's safe limits!

[retired radiologist]

Driving across the cycle lane seems stupid at worst, counter-intuitive at best!

But autonomous cars should find it easier than wetware drivers to comply with disparate rules between jurisdictions.

They have GPS to tell them where they are, and could be programmed with different subroutines for different manoeuvres in different areas without difficulty.

All the E&E departments I've worked in have ring cutters available, which have a blunt bit to slide under the ring between gold and flesh, and a rotary cutter turned by hand to cut through the ring. I'm sure one of these would have worked. Possibly for a thick ring two cuts 180 deg apart would have been required.

The IoT needs two major elements to make it work and become successful.

First is to ensure security - no backdoors from the 'fridge to your network.

Second is to have standardisation of communications between all devices - regardless of manufacturer.

These will need to be agreed like the internet protocols across the interweb, and could be similar to, for example, HL7, XDS and DICOM in the medical field.

That way, users wouldn't be locked in to one supplier [or conversely have different little installations that run independently of each other] and also if a supplier went bust, the kit could still be used with another controller or app.

key taken away from both sides so no one can gain access or exit the door with-out a key.

And no one can open the door when the house burns down, either.

I'd rather be burgled than roasted - though neither would be the ideal outcome. My locks have thumbturns on the inside, so anyone trapped can get out.

Why should we bend over backwards to accommodate people who have already broken the law by failing to register in the first place?

1984

Bring on the telescreen - but in a rather more sophisticated form.

Anonymous?

Not by the time Google has compared the medical data with the vast amount of data it holds on individuals from searches, gmail, social media etc.

It'll have your name, address, phone number and more attached to the medical record in the shake of a lamb's tail.

New systems

There are new systems coming on line so probably best to opt out again. At least 2 Reed codes are needed now, compared with only one previously..

More info here: http://www.nhsdatasharing.info/

William

Another project to sign off without end-to-end testing?

"The otherwise run-of-the-mill assault consisted of"

Shouldn't that be "The otherwise run-of-the-mAll assault consisted of"?

Bosch may be at the root of it. They provided the software for VW, and allegedly have a cheat code in their vacuum cleaners such that power consumption is increased when the bag fills up, to maintain suction, despite renergy tests requiring the low level that only actually works when the bag is empty.

Bosch were also strong lobbyists for the EU test regime to be performed with an empty bag, going against the rea.-world usage.

Dyson is livid - to the extent of taking them to court!

Several years ago my hospital trust had an enterprise-wide phone system with secure SMS and voicemail. It was easy to use [so people used it] and secure.

That is the type of system to use, not personal phones. ICO will fine heavily for data breaches involving the use of personal phones.

Ex Caldicott Guardian.

More nostalgia

Our department had one of these, and i have fond memories of creating dbase databases from the 'dot' prompt.

Sunday drivers

I think they will cause more frustration to drivers in cars around them, as early reports seem to indicate they are very cautious and slow/stop at junctions even though they can be seen to be clear. Anyone attempting to make good but safe progress will soon be shouting at them.

Perhaps they should have grey-haired dummies wearing trilbies and smoking a pipe [e-pipe, of course] to warn following drivers?

Snag is, you have to enter your PIN every time you use it!

I reckon they just closed the windows, then re-opened them.

Limitations required

The last time this was mooted, it seemed that everyone from the council parking attendant upwards would be given access to these data.

It would probably be justifiable if access were limited SOLELY to the intelligence services.

Clean-up

I though JIF was now called CIF !

But as it's a contraction of 'Graphics', surely it should retain the hard G in the abbreviation? Or should graphics have a soft G [danger of confusion with reticulated-skinned creatures with long necks!