Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Medicine's Machine Learning Problem

bostonreview.net

Post reply on HN
Print view View original

21–30 of 112 posts

Re: Medicine's Machine Learning Problem

#21

Post by dnautics »

usrnm wrote: post #18
dontreact wrote: post #3

It's a hard problem to work around which is rooted in the data available. I published this paper while I was at Google: https://www.nature.com/articles/s41591-019-0447-x The only data we were able to get at the time was mostly white patients. We talked to many hospitals but many were/are reluctant to share anonymized data for research. I'm not at Google so I'm not sure the status of the project now, but there was a r…

Honest question: does it really matter for lung cancer? Is there much difference between races in this particular field?

How would you know without the data? There are plenty of medical conditions with wildly divergent rates and pathophysiologies based on human genetics.

Re: Medicine's Machine Learning Problem

#22

Post by sungam »

gdebel wrote: post #14

I'm an eye surgeon and self-taught machine learning practitioner, I started to learn Python in 2016 when the deep learning hype was at his highest. After 3 years of research, playing with datasets, extracting and cleaning data from EMR and from different machines, I not sure that the biggest problem with the so-called "AI" is the inequalities that it can induce ; it is rather, is it useful at all ? This is a little b…

I am a dermatologist, AI researcher and co-founder of an AI startup (skinsmart.ai) and I would agree with you regarding the utility of AI in making an accurate diagnosis of melanoma. I don't think it has a significant role it play in the Dermatology clinic for this application. However, I am very optimistic about the potential for AI to help in the triage of patients referred to dermatology by non-specialists. For this application you are not trying to diagnose melanoma but instead aiming to diagnose - with a high degree of accuracy benign lesions that do not need review in the Dermatology clinic.

Re: Medicine's Machine Learning Problem

#23

Post by krcz »

gdebel wrote: post #14

I'm an eye surgeon and self-taught machine learning practitioner, I started to learn Python in 2016 when the deep learning hype was at his highest. After 3 years of research, playing with datasets, extracting and cleaning data from EMR and from different machines, I not sure that the biggest problem with the so-called "AI" is the inequalities that it can induce ; it is rather, is it useful at all ? This is a little b…

While I don't think AI should replace humans in describing medical images, it can be used to check if they might have missed something. Such AI-based description should be provided only after the human finishes analyzing the image, to avoid lazy technicians just copying algorithmic output. The goal doesn't have to be increasing accuracy and not doing biopsies, it might be reducing number of false negatives.

Re: Medicine's Machine Learning Problem

#24

Post by hobom »

This includes some true things about data collection issues, but I cannot agree with the main thesis that ML algorithms are about power. If anything, they shift power to the patients because now the decisions can be checked and questioned at many levels. An algorithm will not send you away because it is to tired too take your complaints seriously.

So, using algorithmic decision making should be as bias free as possible, but there is no way that across the board they will be more biased than humans are now. If people care about marginalized communities, they should push with everything they got for, not oppose, ML decision making.

Re: Medicine's Machine Learning Problem

#25

Post by lhomdee »

dontreact wrote: post #3

It's a hard problem to work around which is rooted in the data available. I published this paper while I was at Google: https://www.nature.com/articles/s41591-019-0447-x The only data we were able to get at the time was mostly white patients. We talked to many hospitals but many were/are reluctant to share anonymized data for research. I'm not at Google so I'm not sure the status of the project now, but there was a r…

> Fundamentally, it seems to me like there just aren't as many lung cancer screening scans out there for non-white patients as there are for white patients.

Just to qualify, you mean for the USA alone? It seems to me that part of the challenge is recognizing that the research needs to take place beyond just Western countries, or acknowledging it where such research is already occurring. Understandably many people would not be so comfortable with Google accessing patient data from around the world, so the next challenge is how diverse and global data can be protected so that important medical research can take place without any compromise of privacy.

The challenge is hard but surely not impossible, as this was the approach taken by the AstraZeneca-Oxford (and others) which conducted part of its covid vaccine trials in South Africa to test efficacy on non-white populations.

Re: Medicine's Machine Learning Problem

#26

Post by jsinai »

epmaybe wrote: post #13
dontreact wrote: post #3

It's a hard problem to work around which is rooted in the data available. I published this paper while I was at Google: https://www.nature.com/articles/s41591-019-0447-x The only data we were able to get at the time was mostly white patients. We talked to many hospitals but many were/are reluctant to share anonymized data for research. I'm not at Google so I'm not sure the status of the project now, but there was a r…

I certainly see and empathize where you are coming from. However, I would like to add that it kind of makes sense that you’d have more white people with scans available. Focusing on the USA for a second (and note that this likely applies elsewhere too, since screening programs are really only in full force in developed countries which, surprise surprise, are predominantly white). Non white patients don’t get screened…

> Non white patients are inherently fewer than white patients

Look at global population statistics. While there are no official global figures for ethnicity, we can make some simple inferences based on continental distribution [1]:

North America + Europe combined (17.19%) is barely as much as Africa (17.2%), and this is ignoring the fact that a good part of the North American population is non-white. There is nothing "inherent" about there being less non-white patients. The issue is inbalanced access to health care and screening programmes, but that is not inherent.

This is without even mentioning that Asia accounts for almost 60% of the global population.

[1] https://en.wikipedia.org/wiki/Demographics_of_the_world#2020...

Re: Medicine's Machine Learning Problem

#27

Post by DoreenMichele »

dontreact wrote: post #3

It's a hard problem to work around which is rooted in the data available. I published this paper while I was at Google: https://www.nature.com/articles/s41591-019-0447-x The only data we were able to get at the time was mostly white patients. We talked to many hospitals but many were/are reluctant to share anonymized data for research. I'm not at Google so I'm not sure the status of the project now, but there was a r…

How do we improve on the situation?

Given economic realities and racist history (consider what happened in Tuskegee as one example), in the US you would need to provide free screenings to poor people under circumstances that convinced people of color they can trust you while signing the documents to let you have their data.

This is a fairly high bar to meet and one most studies are probably making zero effort to really meet.

I'm part Cherokee and I follow a lot of Natives on Twitter due to sincere and legitimate interest in my Native heritage, but the world deems me to be a White woman so I am sometimes met with hostility simply for trying to talk with Native people while looking too White to be trustworthy. Prior positive engagement with specific individuals seems to carry little weight and be rapidly forgotten. The slightest misstep and, welp, "she's an evil White bitch, here to fuck over the Natives -- like they always are!"

I'm not blaming people of color for feeling that way. I'm just saying that's the reality you are up against.

As someone who spent some years homeless and got a fair amount of "help" offered of the "God, you clearly are an idiot causing your own problems and need a swift kick in the teeth as part of my so-called help" variety, I really sympathize with such reactions.

White people often have little to no understanding of the lives of people of color and little to no desire to try to really understand because really understanding it involves understanding systemic racism in a way that tends to make Whites very uncomfortable. It veers uncomfortably close to self-accusation to honestly try to see how the world is experienced by such people.

Re: Medicine's Machine Learning Problem

#28

Post by twic »

ps2fats wrote: post #17
gdebel wrote: post #14

I'm an eye surgeon and self-taught machine learning practitioner, I started to learn Python in 2016 when the deep learning hype was at his highest. After 3 years of research, playing with datasets, extracting and cleaning data from EMR and from different machines, I not sure that the biggest problem with the so-called "AI" is the inequalities that it can induce ; it is rather, is it useful at all ? This is a little b…

May I suggest, in response to your sentiment that applications of AI to medicine are lacking, is that you are seeing applications replace current medical practices. An AI diagnosis of a medical image seems redundant indeed, however in this situation a patient has seen a doctor out of complaints and has been sent to the radiologist for further investigation. This medical practice is reactionary, and suspicions are alr…

Sounds good, doesn't it? But you have to have a really, really low false positive rate for this to work out. This is already a problem with mammography:

https://www.cochrane.dk/news/new-study-finds-breast-cancer-s...

Re: Medicine's Machine Learning Problem

#29

Post by refurb »

gdebel wrote: post #14

I'm an eye surgeon and self-taught machine learning practitioner, I started to learn Python in 2016 when the deep learning hype was at his highest. After 3 years of research, playing with datasets, extracting and cleaning data from EMR and from different machines, I not sure that the biggest problem with the so-called "AI" is the inequalities that it can induce ; it is rather, is it useful at all ? This is a little b…

First, it took me a very long time to really, fully get that AI is not fundamentally different from a simple linear regression (infering a rule from data).

I had a similar revelation. I sat through an AI for health presentation and basically asked “ok, so you take a data set then try and find a set of rules that accurately describes...like a linear regression?”

As you said, it’s more sophisticated than that, but in essence, yes, it’s fitting a curve to data.

Re: Medicine's Machine Learning Problem

#30

Post by rscho »

The current effort towards medical ai is heading in the wrong direction: were trying to make machines adapt to the field while we should be trying to adapt the field to make it available to machine-aided reasoning. Problem: almost nobody understands both medicine and machines well enough to bridge the abysmal communication gap separating AI/CS and medical professionals.
Post reply on HN