[Will Hunt]

On the data sharing thing, I'm sure I've missed something but I kind of don't mind. I can see the obvious advantages for research and for making the NHS function better. If the data gets sold to a private company I'm fairly nonplussed provided that the data is anonymised. That's a big if, and I don't really trust the government to get that right, but I don't know whether the risk of unanonymised data getting out is worth limiting the advantages of having it shared.

Explain to me like I'm five.

[Offwidth]

They hope people need to think like five year olds.

The issues around governance, democracy, contracts for mates and data security have been warned about for many months.

https://www.digitalhealth.net/2020/11/data-contracts-with-palantir-risk-undermining-core-values-of-nhs/

https://www.opendemocracy.net/en/ournhs/why-were-suing-over-the-23m-nhs-data-deal-with-palantir/

Plus warnings that the data isn't fully anonymous.

https://theconversation.com/your-nhs-data-is-completely-anonymous-until-it-isnt-22924

Which are turning out to be true enough for the government to try and hide the issues

https://www.opendemocracy.net/en/ournhs/fresh-concerns-over-privacy-and-profit-nhs-covid-data-deals/

I speak as someone who massively supports the medical benefits of an anonymous NHS database used under good governance. When we are dealing with known bad actors like Palantir under special covid regulations that bypass democratic scrutiny and normal civil service checks this could be an utter disaster.

[Will Hunt]

OK, good links.

Playing devil's advocate: is it naive to think that a big data company actually needs your medical history to figure out your medical history? Sure, it would help, but most people already give up a huge amount of information about themselves anyway through using Google/Facebook/virtually anything online. Remember the Cambridge Analytica story where they could basically figure out exactly who you were if you had liked Wu-Tang Clan? Even if you think that your data has only been handled by companies that you trust, what makes you think those companies haven't been hacked? What's the expression? If you think you've been hacked then you've been hacked; if you don't think you've been hacked then you've been hacked but just don't realise it.

Also, reading that third link, is it not futile to say that one wants to have a central NHS database and all the medical benefits that go with it AND to have anonymity and privacy assured, since any useful database will not be anonymous.

I wonder whether we ought to just accept that, as computing power increases, everything about us will be known and we should concentrate instead on legislating against that knowledge being used against us. i.e. we don't insist that interviews be blind, but it isn't lawful to reject a candidate on the basis of their sex.

[Bonjoy]

I wonder whether we ought to just accept that, as computing power increases, everything about us will be known and we should concentrate instead on legislating against that knowledge being used against us. i.e. we don't insist that interviews be blind, but it isn't lawful to reject a candidate on the basis of their sex.

Because an interviewer will always be fully conscious of the reasons they didn't hire someone, and honest in how they report them 

[TobyD]

OK, good links.

Playing devil's advocate: is it naive to think that a big data company actually needs your medical history to figure out your medical history? Sure, it would help, but most people already give up a huge amount of information about themselves anyway through using Google/Facebook/virtually anything online. Remember the Cambridge Analytica story where they could basically figure out exactly who you were if you had liked Wu-Tang Clan? Even if you think that your data has only been handled by companies that you trust, what makes you think those companies haven't been hacked? What's the expression? If you think you've been hacked then you've been hacked; if you don't think you've been hacked then you've been hacked but just don't realise it.

Also, reading that third link, is it not futile to say that one wants to have a central NHS database and all the medical benefits that go with it AND to have anonymity and privacy assured, since any useful database will not be anonymous.

I wonder whether we ought to just accept that, as computing power increases, everything about us will be known and we should concentrate instead on legislating against that knowledge being used against us. i.e. we don't insist that interviews be blind, but it isn't lawful to reject a candidate on the basis of their sex.

Although I agree that being too protective of data is pointless as more or less everyone uses Facebook, WhatsApp, Google, Apple or something like that; Dido Harding has a proven record of losing thousands of data records and then lying about it repeatedly at Talk Talk. My brother interviewed her shortly after it and she lied about it to his face. (He's a business journalist).
I'd rather that unknown corporations didn't know if I had a terminal disease or a history of severe depression for example (both hypothetical incidentally). I'm more worried about pure incompetence than malign intent to be honest.

[Bradders]

OK, good links.

Playing devil's advocate: is it naive to think that a big data company actually needs your medical history to figure out your medical history? Sure, it would help, but most people already give up a huge amount of information about themselves anyway through using Google/Facebook/virtually anything online. Remember the Cambridge Analytica story where they could basically figure out exactly who you were if you had liked Wu-Tang Clan? Even if you think that your data has only been handled by companies that you trust, what makes you think those companies haven't been hacked? What's the expression? If you think you've been hacked then you've been hacked; if you don't think you've been hacked then you've been hacked but just don't realise it.

Also, reading that third link, is it not futile to say that one wants to have a central NHS database and all the medical benefits that go with it AND to have anonymity and privacy assured, since any useful database will not be anonymous.

I wonder whether we ought to just accept that, as computing power increases, everything about us will be known and we should concentrate instead on legislating against that knowledge being used against us. i.e. we don't insist that interviews be blind, but it isn't lawful to reject a candidate on the basis of their sex.

Although I agree that being too protective of data is pointless as more or less everyone uses Facebook, WhatsApp, Google, Apple or something like that; Dido Harding has a proven record of losing thousands of data records and then lying about it repeatedly at Talk Talk. My brother interviewed her shortly after it and she lied about it to his face. (He's a business journalist).
I'd rather that unknown corporations didn't know if I had a terminal disease or a history of severe depression for example (both hypothetical incidentally). I'm more worried about pure incompetence than malign intent to be honest.

I just can't get excited about this. Why does it matter that any given organisation could know my medical history?

[ali k]

I just can't get excited about this. Why does it matter that any given organisation could know my medical history?

I’m no data expert but I’m led to believe it wouldn’t be too difficult to de-anonymise the records even now, let alone in 10, 15, 20 yrs time. Adding in the not-uncommon data breaches (both malicious and accidental) personally it makes me not too keen on third parties holding my medical records. Once data is out there it tends to have a habit of being sold or passed on to other interested parties - any one of which could then be subject to a data breach.

If you’re happy for your medical records to be out there that should be up to you. But I don’t think it’s on for the govt to presume everyone is happy for their records to be scraped and the default be to have to opt out. There’s already enough hesitancy among a lot of people in even going to see the GP without adding in the fear of, say, your employer finding out you’ve got 7 testicles and a gambling addiction.

[Bradders]

That still doesn't answer the question of why it matters that anyone might have medical history data? What could anyone possibly do with it that makes it a bad thing for it to be shared?

[mrjonathanr]

Buy it and pick which parts of the NHS they are going after as a private medical business. The tentacles medical businesses are far too interwoven into it already and trade deals with the US will increase access. I certainly don’t think that’s far fetched, looking at the shower in power.

[ali k]

That still doesn't answer the question of why it matters that anyone might have medical history data? What could anyone possibly do with it that makes it a bad thing for it to be shared?

You mean as in worst case scenarios if the data was freely available online due to a data breach and de-anonymised? Just off the top of my head…employers, insurance companies, mortgage lenders discriminating against people with certain conditions. People not going to see the GP for fear of the above. Without even removing anonymity there are countless reasons why private health companies would want this data to seek out the most profitable areas of the NHS and ditch the rest. I’m sure far more knowledgeable and greedy people with a stake in this could come up with a thousand things to do with it to make money.

[mrjonathanr]

It’s pseudonymised data. Not irreversibly anonymised.

[Will Hunt]

It’s pseudonymised data. Not irreversibly anonymised.

Forgive me if I've missed something, but does this not belie how de-anonymisation will work? It won't be that some clever nerd figures out an algorithm that turns your pseudonym back into a name, rather, by a process of elimination they will figure out who you are, since each individual's information will contain clues as to their identity.

So (paraphrasing the 3rd link that OW posted above) they'll get a good idea of your age by when you received certain vaccinations, or maybe they'll be able to lump you into a particular age bracket by what treatment you've received in the past. They'll know roughly where you live because of who your GP is, they'll know your previous address history and roughly when you moved, they'll know who your immediate family is because their movement history will match your own etc etc etc. Pair that up against a load of other available data that's out there and you've matched the medical data to an individual.

The reason that I think the method of de-anonymisation is relevant is that decrypting a pseudonym would be the result of incompetence or maliciousness on the part of an insider, while large-scale de-anonymising of medical data is an inevitability - just a matter of applying the right amount of computing power to it.

[Will Hunt]

I'd rather that unknown corporations didn't know if I had a terminal disease or a history of severe depression for example (both hypothetical incidentally).

I think my point is, what makes you think that this information isn't already known? I'd hazard a guess that Google knows more about my health than my GP does.

I probably will opt out, but I think it's a shame. Giving access to big datasets has potentially huge benefits for medical research. Anecdotally, I remember hearing a researcher speak who had led a study that looked at the incidence of SIDS (cot death). They noticed that there was a significantly lower incidence of this in Birmingham. After investigating further they found that mothers from the large (if I remember correctly) Indian community were doing as their mothers had taught them and laying their babies to sleep on their backs, which greatly reduces the incidence of SIDS. The advice given to parents was changed and we now have a far lower rate of SIDS.

Take all the GP data in the country and throw some machine learning at it and who knows what you'll find out.

[TobyD]

That still doesn't answer the question of why it matters that anyone might have medical history data? What could anyone possibly do with it that makes it a bad thing for it to be shared?

As Ali said above, employers could refuse job applications based on medical conditions,  insurance companies hiking up premiums,  aggressive sales / advertising ...

[mrjonathanr]

It’s pseudonymised data. Not irreversibly anonymised.

Forgive me if I've missed something, but does this not belie how de-anonymisation will work?

https://www.grcworldforums.com/systems-security/data-masking-anonymisation-or-pseudonymisation/12.article

[TobyD]

Reading this news:
BBC News - Russian jets and ships shadow British warship
https://www.bbc.co.uk/news/world-europe-57587777

Isn't it a coincidence that the Royal Navy seems to come out shortly before elections which matter to the government? I'm sure this is nothing but a pea brained conspiracy theory,  but it does seem to happen.  They probably aren't actually competent enough to do something like that.

[Oldmanmatt]

As regards insurance companies
, lenders etc. You will be asked to make a health declaration (this is also true for many professions and vocations, many already require medical examinations anyway) and, should it become apparent that you were hiding a disqualifying condition, you would be denied insurance payout, or dismissed from your employment.
If you are in employment and you become a burden to your employer through extended absence, inability to perform etc, they will find a way to get rid of you. Interviews are not blind (whether they should be or not, is a different story), so employers will (and do) assess on several things that they are not legally allowed to. All the law can ever hope to achieve is preventing employers from stating a proscribed reason for rejection or dismissal.

I’m not sure why this data usage should be seen as so ominous. The costs of processing such data, in such volumes, would seem quite prohibitive. So probably a third party investment, with the aim of selling the decrypted data for profit. This would mean that only very insurances or very highly remunerated roles, would be subject to such prior investigation (when I was in the process of joining Lurssen as a senior project manager, I was presented with a four page health declaration, an appointment for a full medical and a contract that required me to refrain from participation in “dangerous sports” such as skiing or rock climbing. I didn’t pursue the application any further).

So, for instance, my mother and father, both cancer survivors, find holiday insurance very hard to obtain or expensive. If they chose to hide their survior status, they could certainly save money and effort, however, should they ever need to actually use their insurance, they might very well not only not receive the cover they need, they might find themselves in some legal hot water, too boot.
 
 

[Oldmanmatt]

Reading this news:
BBC News - Russian jets and ships shadow British warship
https://www.bbc.co.uk/news/world-europe-57587777

Isn't it a coincidence that the Royal Navy seems to come out shortly before elections which matter to the government? I'm sure this is nothing but a pea brained conspiracy theory,  but it does seem to happen.  They probably aren't actually competent enough to do something like that.

It never happened.
The reports arose from some Russian News agencies. MOD has denied it happened. The ~200 crew members all said “uh, what?” When their worried families started bombarding them with phone calls and the MOD asked relatives to stop calling them, too.
We routinely escort “Warships of hostile nations” through places like the Channel, with both ships and aircraft.

Edit.

To clarify. Yes, they heard shots.

The Russian gunnery exercise was scheduled, promulgated in good time through the normal channels and went as described.
It was “purely coincidental” that an RN ship was within earshot (read observation range) of a Russian exercise and they were not in anyway threatened. One paper (the Mail? Don’t recall) even tried to suggest bombs had been dropped in the ships path. It’s just bollocks.

We have always taken quite provocative actions with hostile Navies. I spent several very boring days in total silence, in the run up to Desert Storm, whilst the ship I was on, charged at Russian ships anchored in the Black Sea, cut power and coasted between vessels at unsafe distances, whilst listening intently with some fairly advanced passive sonar. There will have been “diplomatic protests” for sure, just as we return the favour.
That’s not even half the story, either. Vessels are often damaged in these silly games.

[Bradders]

That still doesn't answer the question of why it matters that anyone might have medical history data? What could anyone possibly do with it that makes it a bad thing for it to be shared?

As Ali said above, employers could refuse job applications based on medical conditions,  insurance companies hiking up premiums,  aggressive sales / advertising ...

As Matt pointed out, if you're hiding something of relevance from your insurance company you are committing fraud, and at best will not be able to claim on your policy. If you have previous conditions then you genuinely are more of an insurance risk and the potential cost to the firm is greater, ergo you pay a higher premium. They're not charities!

My point is there seems to be this instinctive ideal that data should be retained and preserved, when data is good and useful and often leads to highly positive outcomes. It strikes me as a sort of double think where people want help with their problems, but don't want to give other people the tools to do something, because there's a microscopic risk of.....what? The potential negative outcomes just don't make any sense in the real world.

[sdm]

As regards insurance companies
, lenders etc. You will be asked to make a health declaration (this is also true for many professions and vocations, many already require medical examinations anyway) and, should it become apparent that you were hiding a disqualifying condition, you would be denied insurance payout, or dismissed from your employment.
If you are in employment and you become a burden to your employer through extended absence, inability to perform etc, they will find a way to get rid of you. Interviews are not blind (whether they should be or not, is a different story), so employers will (and do) assess on several things that they are not legally allowed to. All the law can ever hope to achieve is preventing employers from stating a proscribed reason for rejection or dismissal.

I’m not sure why this data usage should be seen as so ominous. The costs of processing such data, in such volumes, would seem quite prohibitive. So probably a third party investment, with the aim of selling the decrypted data for profit. This would mean that only very insurances or very highly remunerated roles, would be subject to such prior investigation (when I was in the process of joining Lurssen as a senior project manager, I was presented with a four page health declaration, an appointment for a full medical and a contract that required me to refrain from participation in “dangerous sports” such as skiing or rock climbing. I didn’t pursue the application any further).

So, for instance, my mother and father, both cancer survivors, find holiday insurance very hard to obtain or expensive. If they chose to hide their survior status, they could certainly save money and effort, however, should they ever need to actually use their insurance, they might very well not only not receive the cover they need, they might find themselves in some legal hot water, too boot.

There's a big difference between what you already have to declare to insurance companies, employers etc and having private companies being able to purchase your entire medical history (cross referenced with that of all of your relatives, and deanonymised thanks to your social media, google accounts etc).

It wouldn't just be your parents who may have to pay more for their health/travel insurance, it could be you and your children too. Maybe you can't get a mortgage because they deem you to be at a higher risk of death/not being able to work before your mortgage is paid off/in positive equity.

On the other hand, someone who has a family with a relatively clean family health history may find themselves in a position of having to pay much more for a pension annuity or care in their old age because they are likely to live longer.

Employers could overlook someone who has received fertility advice/treatment because they don't want to take someone on who then goes off on maternity/paternity leave. Or they'll just overlook anyone who has a history of mental health issues etc. All of this would be illegal but isn't something you would ever be able to prove.

Paternity and family law could be a huge problem. Cross referencing an entire family history would allow companies to assess whether two people are related.

A reasonable expectation of anonymity is essential to ensure that people seek the help they need. Without that, vulnerable people are likely to avoid getting help for mental or sexual health problems or domestic abuse because of concerns over future blowback.

People in vulnerable positions could be targeted by unscrupulous companies. Alcohol or gambling adverts could be targeted to recovering addicts. Scams, dodgy loans, extreme politics/religions/cults etc could be targeted to people who are desperate and more likely to fall for them.

There are too many situations where a person's health history could be used against them and, with the records of big data companies, I don't see any reason to think that this won't happen. With sufficient legal safeguards, it is possible to get the research benefits of a centralised system while avoiding many of the concerns above. Without those safeguards, I felt I had no choice than to opt out. Your medical data is guaranteed to end up in the wrong hands, either by design or by poor security.

[ali k]

My point is there seems to be this instinctive ideal that data should be retained and preserved, when data is good and useful and often leads to highly positive outcomes.

But it can also have some highly negative outcomes which might be hard to appreciate in the early days. When facebook was born and it was all about connecting with friends around the world (positive outcome), did anyone really imagine that 20yrs later it would become a tool that could be used to manipulate elections by micro-targeting different people with different adverts based on their prejudices and vulnerabilities using the data they had accumulated (I would argue disastrously negative outcome). Same could be said for a lot of social media companies, which were started with only good intentions but are now having to firefight to try and contain the negative outcomes.

there's a microscopic risk of.....what? The potential negative outcomes just don't make any sense in the real world.

Maybe it's hard to perceive any negatives at this point in time if you're confident it will only make health research and outcomes better. But personally I think it's naive to think that the way insurance companies (or employers, mortgage lenders and the rest) go about their actuarial activities at this point in time might not be affected in a negative way once they have access to reams and reams of medical data. See sdm's post (cross-posted).

[petejh]

Anyone who finds it hard to imagine the potential negative consequences of having their NHS data shared with commercial interests would be wise to get talking to data scientists from one of the large American insurance companies. Allstate, for example, run a data analysis operation from their offices in Northern Ireland (low wages for high skills). My ex worked as a data scientist and I learned about some of Allstate's methods of sifting terabytes of 'anonymised' data to identify and target individuals in the US. Completely scary the length these organisations go to to tilt the market slightly in their favour, so that they can either sell more products, assume less risk, or pay people less. While the data collected from the Northern Ireland offices is used to target individuals in the US and Canadian insurance markets (along with god knows what else..), you can bet the UK won't be untouched for much longer by operations like these, if it isn't already. As sdm correctly says the only sensible option is to complete the NHS online type-1 opt out. Which is what I've done.   

[Will Hunt]

The costs of processing such data, in such volumes, would seem quite prohibitive.

It might seem like this is the case now, but as Pete points out, it isn't. And even if it wasn't, to assume that it would remain the case would be naïve.

Insurers might ask you now for your medical history. How far away are we from a world where an insurer can make a reasonable stab at guessing your medical future? They could easily have your medical history and that of your family, and from that they can guess whether you're likely to get cancer in the next 5 years. You might find yourself being denied insurance (and thus the right to travel safely etc) based on a possible future illness. Hold onto your tin foil hats, but if we continue to make advances in genome sequencing then who's to say that an insurer can't analyse this and immediately deem you unsuitable for insurance? Consider this in a healthcare system not free-at-the-point-of-need.

As Ali pointed out, nobody would have thought 15 years ago that Facebook could be used to deliver personalised political messages based on a person's accurately inferred values. You don't have to mention politics on Facebook for somebody to figure out exactly what you think about politics.

Worth ten minutes of your time if you think this is all sci-fi nonsense:

[Dac]

NHS data can already be used for research, for free; it it just a matter of demonstrating that all data protection, anonymity and ethical considerations will be met.

So if these companies only intend to do the same then why are they willing to pay for the data?

[Will Hunt]

the only sensible option is to complete the NHS online type-1 opt out. Which is what I've done.   

Is this all you have to do for the opt out? I was thinking I had to send a form to my GP?