The Computer Will See You Now

[Clitheroekid]

A fascinating programme from the Horizon stable about the use of AI in medical diagnosis - https://www.bbc.co.uk/iplayer/episode/b ... ee-you-now

It throws up some interesting ethical dilemmas. In London the company featured, Babylon, has set up a GP service called GP At Hand. There's an explanation of how it works here - https://www.england.nhs.uk/london/our-w ... act-sheet/

It sounds great, and is something I'd certainly be interested in if I lived in the catchment area. However, what's happened is that typical users tend to be young, well-educated and medically aware people who are probably the least likely to need GP services.

Every time one of them decamps to GPAH their previous GP loses the capitation fee that they generated. So GP practices are losing the `profitable' patients to GPAH, and are facing not only the resultant loss of income but the additional burden of an increasing proportion of patients that need a lot of care.

It's an interesting ethical dilemma for the NHS. On the face of it this is a very clever technological innovation that has the potential to increase substantially the efficient use of NHS resources. On the other hand it has the potential to destroy established GP practices.

There was also a very interesting experiment in which Babylon set up a contest between independent GP's and its own AI systems to diagnose 100 different patients. I won't reveal the results in case anyone wants to watch it, but they were very interesting.

My personal view is that it’s almost inevitable that an AI diagnosis will be more accurate than the average GP diagnosis, if not now then in the near future, simply because AI is constantly improving whilst GP's aren't.

The programme commentary is by a woman who I felt to be somewhat biased against AI, but I didn't realise until the closing credits that she was ... Doctor Hannah Fry.

I would imagine it must be very difficult for Babylon to obtain any evidence that is completely unbiased simply because the people who can provide the best independent evaluation are GP’s, who are almost inevitably going to be heavily biased against AI.

I would expect the large majority of GP's to believe that they are better than a machine, just as any other professional would. But as intelligent people they must also accept the reality that it's simply a matter of time before their diagnostic skills will be inferior to that of a machine.

Professionals in every field - including law - are facing the same phenomenon, and there's no doubt that it's very unsettling for people who have always seen themselves (and generally been seen by others) as an elite to realise that in some ways they’re like the craft printers of the 1980's, possessing hard won and highly remunerative skills that may in the uncomfortably near future become obsolete.

[mc2fool]

The programme commentary is by a woman who I felt to be somewhat biased against AI, but I didn't realise until the closing credits that she was ... Doctor Hannah Fry.

That's Dr Hannah Fry, Associate Professor in the Mathematics of Cities at the Centre for Advanced Spatial Analysis at UCL. Her doctorate (PhD) was in fluid dynamics. http://www.hannahfry.co.uk/

[superFoolish]

Hi CK; thanks for your post, I'll try to track down that programme (I'm not in the UK); it sounds very interesting.

...On the other hand it has the potential to destroy established GP practices.

When I think of this, I think of all the people who were going to be put out of work following the invention of tractors, and all those who were going to be put out of work by computers (and self-service petrol stations, supermarket checkouts, etc). Exactly the opposite happened (more jobs were created)!

My wife is a manager of theatre services in a large, very modern hospital, and has had some involvement in the implementation of a cutting-edge (pun intended) robot that is used for prostate surgery. Apart from reducing surgery time, use of the robot reduces a patient's hospital stay from 3-5 days to 23 hours (due to accuracy of surgery and minimal blood-loss) leading to savings of millions of dollars per year. One might worry about the loss of jobs for the associated nursing staff, but the facts are that those resources (human and financial) have been redirected elsewhere, to the benefit of other patients, many more patients are able to receive treatment for prostate cancer, and their recovery rates are far improved, leading to future savings (e.g. treatment not required for terminally-ill cancer sufferers).

I don't have the evidence to suggest that will be the case with the introduction of AI, but it's worth bearing in mind that it doesn't necessarily mean loss of jobs (or even loss of important skills), or loss of important health resources for some demographics.

I have not done any formal research into this, but we know that few GPs currently have the time to sit and listen to their patients (they have very limited time available; minutes per patient), which is an important part of treatment (certainly with so many mental-health issues these days). The implementation of the AI system that you mentioned may release GPs and other health-staff to spend time on other areas of healthcare.

There is no evidence that homeopathy has any direct beneficial effect on patients, but evidence does show that patients who attend 'well-run' homeopathy clinics do see health benefits. It is hypothesised (because evidence is thin-on-the-ground) that the health benefits of homeopathy not only come from the placebo effect, but also from the TLC given by the clinics (which charge a lot of money for phials of water, and are thus well-funded). If the resources of 'established GPs' are directed towards 'TLC', that may be a benefit in itself.

I'll be interested to review my above comments after watching the programme!

[Dod101]

That's Dr Hannah Fry, Associate Professor in the Mathematics of Cities at the Centre for Advanced Spatial Analysis at UCL. Her doctorate (PhD) was in fluid dynamics. http://www.hannahfry.co.uk/

That makes her a clever cookie in my books. Ever tried studying fluid dynamics?

As to the whole concept, the trouble is CK, that you say it may 'destroy established GP practices'. These are being destroyed anyway by the fact that there are 'way insufficient GPs in the system and they are resigning/retiring faster than the replacements can be found. This sounds like a great move forward and is the sort of thing that is badly needed in the NHS.

What superfoolish has your homeopathy comments got to do with it?

Dod

[UncleEbenezer]

The programme commentary is by a woman who I felt to be somewhat biased against AI, but I didn't realise until the closing credits that she was ... Doctor Hannah Fry.

That's Dr Hannah Fry, Associate Professor in the Mathematics of Cities at the Centre for Advanced Spatial Analysis at UCL. Her doctorate (PhD) was in fluid dynamics. http://www.hannahfry.co.uk/

There may be an element of BBC party line in there.

A scientist who was really enthusiastic about AI might get a platform as interviewee or panel member, but isn't going to get invited to present the whole programme. Fry is a BBC regular favourite, meaning she's media-trained and knows when to keep a dissenting line to herself (if applicable).

[redsturgeon]

Very interesting programme indeed and thanks for the heads up. Mrs RS and I run a healthcare consultancy and we touch on the edges of this stuff.

The interesting thing for me is the collision between the world of tech start ups vs the world of peer reviewed evidence based medicine. As always the legislation has to run to try to catch up with the tech companies in order to keep things safe and avoid the law of unintended consequences from the promises of new tech which can solve problems in one area and save costs while creating problems in another. As someone in the programme mentioned, are the cost savings merely going to end up in the pockets of the tech companies while healthcare spend goes down?

AI technology undoubtedly has a lot to offer medicine and healthcare but as usual the fight between the vested interests on both sides, BMA vs big tech will probably end up poorly managed by the law makers such that the opportunity is not maximised for the benefit of the patient.

John

[Dod101]

What does a healthcare consultancy do?

Dod

[redsturgeon]

What does a healthcare consultancy do?

Dod

I'd have to charge you a fee to tell you.

We work with various organisations within the healthcare universe. That could be hospitals, pharmacies, Local Pharmacy Committees or Pharmaceutical Companies. We advise on marketing strategies, facilitate meetings, train staff, provide distance learning, individual coaching or just act as an expert resource on current legislation or trends within the market.

John

[Dod101]

Interesting thanks.

Dod

[UncleIan]

There may be an element of BBC party line in there.

A scientist who was really enthusiastic about AI might get a platform as interviewee or panel member, but isn't going to get invited to present the whole programme. Fry is a BBC regular favourite, meaning she's media-trained and knows when to keep a dissenting line to herself (if applicable).

I don't quite get what you mean. Are you saying the BBC has an anti science bias? Or they like their presenters to be decent presenters? What would she be dissenting against?

[redsturgeon]

The BBC has a duty to be objective and unbiased in their reporting...I guess that covers documentaries too. So an enthusiastic supporter of AI or an enthusiastic opponent would not get the job of presenting such a programme. The idea is to have a neutral presenter and have interviews with others both pro and anti.

EG. In the is programme we saw the obviously enthusiastic techies and also got the counter view from the doctors representatives.

If Hannah Fry was an enthusiast for AI (which she probably is) then it is her job to put that bias to one side for the programme.

FWIW I thought the programme was reasonably balanced.

John

[Slarti]

As to the whole concept, the trouble is CK, that you say it may 'destroy established GP practices'. These are being destroyed anyway by the fact that there are 'way insufficient GPs in the system and they are resigning/retiring faster than the replacements can be found.

A situation that is only going to get worse, given that our wonderful chancellor is reducing the funds for doctor and nurse training for next year.

Slarti

[Lootman]

the trouble is CK, that you say it may 'destroy established GP practices'. These are being destroyed anyway by the fact that there are 'way insufficient GPs in the system and they are resigning/retiring faster than the replacements can be found. This sounds like a great move forward and is the sort of thing that is badly needed in the NHS.

But who says the GP system is the only way to handle front-line healthcare?

The US system manages without any concept like GPs. (Except that some HMO's require that you first go through a gatekeeper before you see a specialist). Otherwise a patient can attend a clinic, hospital or book directly with a specialist.

Maybe our GP system needs a shakeup.

[dionaeamuscipula]

the trouble is CK, that you say it may 'destroy established GP practices'. These are being destroyed anyway by the fact that there are 'way insufficient GPs in the system and they are resigning/retiring faster than the replacements can be found. This sounds like a great move forward and is the sort of thing that is badly needed in the NHS.

But who says the GP system is the only way to handle front-line healthcare?

The US system manages without any concept like GPs. (Except that some HMO's require that you first go through a gatekeeper before you see a specialist). Otherwise a patient can attend a clinic, hospital or book directly with a specialist.

Maybe our GP system needs a shakeup.

US health spending is about double per capita versus other wealthy countries:

https://www.healthsystemtracker.org/cha ... -countries

Adults wait longer to see a doctor in the US than in many comparable countries (scroll down to near the bottom):

https://www.healthsystemtracker.org/cha ... t-15-years

DM

[Dod101]

[But who says the GP system is the only way to handle front-line healthcare?

No one is saying that but it is very well entrenched here and before they make changes there I would like to see some changes in the funding of the NHS as a whole.

My experience of US health care is limited to two instances which were both covered by insurance but I did not see them as expensive and the service was great with no delays.

Dod

[simsqu]

That's Dr Hannah Fry, Associate Professor in the Mathematics of Cities at the Centre for Advanced Spatial Analysis at UCL. Her doctorate (PhD) was in fluid dynamics. http://www.hannahfry.co.uk/

That makes her a clever cookie in my books. Ever tried studying fluid dynamics?

Yes

[melonfool]

I saw a GP in the US. They definitely exist - it was on insurance, but it was def a general practitioner, I had a chest infection and got anti biotics, it wasn't a specialist.

Mel

[scotia]

That makes her a clever cookie in my books. Ever tried studying fluid dynamics?

Yes

A tale is told of a lecturer who confused her fluid dynamics class by mistakenly thinking that they were her classical electromagnetism class. The students were accustomed to copying down masses of vector differential equations, and it was well into the lecture before the penny dropped.

[Gengulphus]

That's Dr Hannah Fry, Associate Professor in the Mathematics of Cities at the Centre for Advanced Spatial Analysis at UCL. Her doctorate (PhD) was in fluid dynamics. http://www.hannahfry.co.uk/

That makes her a clever cookie in my books. Ever tried studying fluid dynamics?

You seem to be missing mc2fool's point. From the OP:

The programme commentary is by a woman who I felt to be somewhat biased against AI, but I didn't realise until the closing credits that she was ... Doctor Hannah Fry.

Especially given the emphasis on "Doctor", that clearly implies that she is biased against AI because she is herself a GP - an implication that the fact that her doctorate is not medical obviously refutes. She might still be biased against AI, of course, but if so, her reasons for being so are not that obvious.

Gengulphus

[Dod101]

That's Dr Hannah Fry, Associate Professor in the Mathematics of Cities at the Centre for Advanced Spatial Analysis at UCL. Her doctorate (PhD) was in fluid dynamics. http://www.hannahfry.co.uk/

That makes her a clever cookie in my books. Ever tried studying fluid dynamics?

You seem to be missing mc2fool's point. From the OP:

The programme commentary is by a woman who I felt to be somewhat biased against AI, but I didn't realise until the closing credits that she was ... Doctor Hannah Fry.

Especially given the emphasis on "Doctor", that clearly implies that she is biased against AI because she is herself a GP - an implication that the fact that her doctorate is not medical obviously refutes. She might still be biased against AI, of course, but if so, her reasons for being so are not that obvious.

Gee thanks. I had worked that out, believe it or not, all by myself. It still makes her a clever cookie though, at least in my book.

Dod