NHS and Amazon

rick_chasey

PhilipPirrip wrote:

It's not that long since the government were looking to monetize patient data for the purposes of research - they claimed - until it was revealed several of the interested parties were insurance companies.

Despite claims of anonymization, the potential existed for individual patients to be identified and the costs for, and their access to treatment or insurance affected.

Whilst Amazon could collect data from this in a broad 'Health of the Nation' overview type way it's hard to see how they could collect the right types of personal and health data that could be monetized outside of their own interest in offering tailored promotion of their own goods and services.

I think the general population as a whole, and I certainly include myself in this, are very behind the curve around data and what firms and govt's do and don't without them.

I don't think most people in any authority to have an impact on this really know either. Politicians certainly don't, understandably.

The world of data regulation will look very very different in a decade.

rjsterry

awavey wrote:

PhilipPirrip wrote:

Despite claims of anonymization, the potential existed for individual patients to be identified and the costs for, and their access to treatment or insurance affected.

no the potential existed for certain medical conditions to be data mined, such that when individuals requesting insurance policies from these companies and meeting certain data markers, could then be given entirely tailored insurance based on the outcomes from the anonymised data research.

The endpoint of this is not so much insurance - which implies some risk taken on by the insurer - as charging people for healthcare on the basis of a statistical profile.

slowbike

Shirley Basso wrote:

Slowbike wrote:

TheBigBean wrote:

Arthur Scrimshaw wrote:

PBlakeney wrote:

Who thought doctors would have been amongst the first victims of AI?

Well I would for one although this probably isn't the mechanism.

AI consistently outperforms medical professionals in diagnosing health issues in trials, won't be long before it's rolled out hopefully. Remote health monitoring and AI are some of the great opportunities to manage the fact that demand for healthcare is outstripping resources at the moment.

Doctors rely on knowledge. Computers are good at that.

But only if the computer understands the question - something that a human can do better (at the moment). It’s quite tricky searching for something you don’t know the name of.

It doesn't need a question. The computer can ask questions to the patient about symptoms and narrow it down like a decision tree. With remarkable accuracy.

Yes - it can - but your common search engine doesn't do this - it returns results from keywords/phrases you put in ... so you need to know what to ask - or know how to ask to find out what you need to ask. I've come across this in my work world where I've not known the correct term for something - trying to establish what I'm trying to search for is the first task.
If I was talking to someone else in my field then they (hopefully) would know the correct term and direct me accordingly - yes, you can code a computer to do the same, but it has to first establish what you're trying to achieve.

shirley_basso

Slowbike wrote:

Shirley Basso wrote:

Slowbike wrote:

TheBigBean wrote:

Arthur Scrimshaw wrote:

PBlakeney wrote:

Who thought doctors would have been amongst the first victims of AI?

Well I would for one although this probably isn't the mechanism.

AI consistently outperforms medical professionals in diagnosing health issues in trials, won't be long before it's rolled out hopefully. Remote health monitoring and AI are some of the great opportunities to manage the fact that demand for healthcare is outstripping resources at the moment.

Doctors rely on knowledge. Computers are good at that.

But only if the computer understands the question - something that a human can do better (at the moment). It’s quite tricky searching for something you don’t know the name of.

It doesn't need a question. The computer can ask questions to the patient about symptoms and narrow it down like a decision tree. With remarkable accuracy.

Yes - it can - but your common search engine doesn't do this - it returns results from keywords/phrases you put in ... so you need to know what to ask - or know how to ask to find out what you need to ask. I've come across this in my work world where I've not known the correct term for something - trying to establish what I'm trying to search for is the first task.
If I was talking to someone else in my field then they (hopefully) would know the correct term and direct me accordingly - yes, you can code a computer to do the same, but it has to first establish what you're trying to achieve.

Well in this case it's clearly trying to find out what's wrong with you.

Tell Alexa: I'm unwell
Alexa goes into medical diagnosis mode and asks you questions like a doctor would.
Issues likely diagnosis. Sends you do specialist to confirm. Treatment issued. Job done.

TheBigBean

Slowbike wrote:

Shirley Basso wrote:

Slowbike wrote:

TheBigBean wrote:

Arthur Scrimshaw wrote:

PBlakeney wrote:

Who thought doctors would have been amongst the first victims of AI?

Well I would for one although this probably isn't the mechanism.

AI consistently outperforms medical professionals in diagnosing health issues in trials, won't be long before it's rolled out hopefully. Remote health monitoring and AI are some of the great opportunities to manage the fact that demand for healthcare is outstripping resources at the moment.

Doctors rely on knowledge. Computers are good at that.

But only if the computer understands the question - something that a human can do better (at the moment). It’s quite tricky searching for something you don’t know the name of.

It doesn't need a question. The computer can ask questions to the patient about symptoms and narrow it down like a decision tree. With remarkable accuracy.

Yes - it can - but your common search engine doesn't do this - it returns results from keywords/phrases you put in ... so you need to know what to ask - or know how to ask to find out what you need to ask. I've come across this in my work world where I've not known the correct term for something - trying to establish what I'm trying to search for is the first task.
If I was talking to someone else in my field then they (hopefully) would know the correct term and direct me accordingly - yes, you can code a computer to do the same, but it has to first establish what you're trying to achieve.

I suspect your computer doctor would still need a vaguely skilled operator to do the translation between the patient and its data. That level of skill would be far below a doctor though.

rick_chasey

rjsterry wrote:

awavey wrote:

PhilipPirrip wrote:

Despite claims of anonymization, the potential existed for individual patients to be identified and the costs for, and their access to treatment or insurance affected.

no the potential existed for certain medical conditions to be data mined, such that when individuals requesting insurance policies from these companies and meeting certain data markers, could then be given entirely tailored insurance based on the outcomes from the anonymised data research.

The endpoint of this is not so much insurance - which implies some risk taken on by the insurer - as charging people for healthcare on the basis of a statistical profile.

That depends if the NHS model remains the same, which I think in the long term is unlikely.

I imagine in the long term the UK moving to a more hybrid model where insurance is compulsory, and below a certain income/wealth threshold people are giving subsidised or full insurance, and those above will have to cough up.

Insureres are already slighly breaking the model of insurance through their use of data(whereby, by design, those who do not claim subsidise those who do), to a model whereby higher risk people become uninsurable.

Good article here: https://ftalphaville.ft.com/2016/09/22/ ... urce=ifttt (you need to sign up but this blog is free to read).

here’s only one problem. Personalising insurance contracts to this degree undermines the whole concept of insurance.

Insurance doesn’t really work unless risk is pooled in such a way that good agents pay over the odds to the benefit of the bad ones.

A world where insurance is personally tailored to reflect every individual’s behavioural history is consequently a world wherein unfettered discrimination becomes the celebrated norm, where only the super rich, the super gifted, the super lucky or the genetically well-endowed will ever be allowed to drive fast, eat bad food or to take any risk at all. As a consequence, it’s also a world where the poor or physically under-privileged become uninsurable (unless, of course, they’re prepared to be permanently tracked, controlled and monitored by their data overlords).

joenobody

PhilipPirrip wrote:

Whilst Amazon could collect data from this in a broad 'Health of the Nation' overview type way it's hard to see how they could collect the right types of personal and health data that could be monetized outside of their own interest in offering tailored promotion of their own goods and services.

Data doesn't need to be high quality to be monetised. Or rather, data providers are often happy to sell poor quality data while there are people prepared to buy poor quality data.

angry_bird

bianchimoon wrote:

PhilipPirrip wrote:

I've only glanced at this story but Im perplexed why you think this will be a paid for service.

Essentially, when someone asks Alexa a health related question Alexa will use symptom/diagnosis/treatment info from the NHS sites, the same information that you're already free to read by just going to the websites.

It can't be any worse than calling 111.

Not a lot could be, NHS 111 is for people who haven't got the ability to google/learn about their symptoms and make and educated choice of what to do next.

Googling symptoms, classic. "I've checked Google and it says I've got this". Very easy to read something on the internet, different matter understanding the information and the subtleties about how different signs and symptoms go into making a diagnosis.

arthur_scrimshaw

Shirley Basso wrote:

Slowbike wrote:

TheBigBean wrote:

Arthur Scrimshaw wrote:

PBlakeney wrote:

Who thought doctors would have been amongst the first victims of AI?

Well I would for one although this probably isn't the mechanism.

AI consistently outperforms medical professionals in diagnosing health issues in trials, won't be long before it's rolled out hopefully. Remote health monitoring and AI are some of the great opportunities to manage the fact that demand for healthcare is outstripping resources at the moment.

Doctors rely on knowledge. Computers are good at that.

But only if the computer understands the question - something that a human can do better (at the moment). It’s quite tricky searching for something you don’t know the name of.

It doesn't need a question. The computer can ask questions to the patient about symptoms and narrow it down like a decision tree. With remarkable accuracy.

Also asking questions is only part of the diagnostic process and is a bit subjective. The other elements are largely objective (blood tests, smears etc) - MRI scans etc somewhere in between, where AI scores is assimilating all these threads and pulling them into a diagnosis and doing it in a fraction of the time and with greater success rate than a human.

john80

On a lighter note when you think you have an sti, amazon will be able to sell you the testing kit and then regardless of the result it is the clap medication banner ads for the rest of your life. It will maake lunchtime surfing at work a little bit more risky.