Collating a private database of health history is an extremely dangerous practice and I will just drop this article from rms which has further links to other systems [0].
The article in this post, however, is lacking an explanation of how MIB gathers the medical records. It is therefore unclear if the users gave their consent to share the information even if it was through vague fine print. On the other hand the story would be very different if the data was gathered and shared without consent and I might wonder if a legal case can be made here.
That article has no sense of irony, they don't even consider that someone might not want that. How was my info sent to that MIB group? I don't want that. Can you opt out? How is this even legal?
It's ironic how every now and again I read a piece by RMS that I find exaggerated only to come back to it a few years later just to find he was right. Come on, offering people DNA testing only to sell it to insurance companies? That's a new low to me.
When you apply for life and disability insurance you consent to two things: the insurer may collect medical data about anything you mention on the application AND any or all of that may be shared with MIB. Approval is conditional on that consent and omitting anything on the app can lead to denials in the first two years and is technically insurance fraud.
So basically, if you don’t want your information in MIB don’t apply for insurance that requires a medical history.
Various third parties have real-time access to your hospital admissions and prescriptions, in some cases before your insurer.
As an example, when my wife was admitted to the hospital at 8 weeks for a miscarriage that almost killed her, the level of detail provided to data brokers allowed Enfamil to Fedex a big sample box of formula to us on what would have been the due date of the baby.
Going out on a limb here, but if they are paying the bill, don't you think they have a pretty good idea what things cost? At the end of the day there are not that many things that cost $59,324.49 at Hospital X for example. So with a little deduction you could begin to ballpark the severity of illness
Along with the GDPR, the EU also bought "Right to Explanation" for algorithmic decision making. Basically, a user has the right to ask for an explanation on an algorithmic decision. This regulation has been mostly buried in the deluge of GDPR content but it's an important shift in decision making. There's a nice paper which outlines the implications and HN too had a discussion on it couple of years ago [0, 1].
Yea, I think showing your work is substantially more important when people's lives and finances are on the line as opposed to the thrill of victory for a go master.
If you're making algos for that directly affect another human's basic needs, I think it's really important to be able to back up why the algo did what it did. I feel like you're probably opening yourself up to massive liability (regardless of region your operating in) if you can't explain to a court why your system did X to a person that resulted in them suing your company.
You cannot then use such a network to make a questionable decision. In other words, they better be robust or you might drown in lawsuits or face a big fat one.
Which is the exact result people want. We want explanations that are valid and understandable in relatively plain language, using postulate logic.
It's not actually hard to come up with a valid and understandable explanation from a neural net, or from any other black box decision maker as long as the decision maker is consistent.
Let's take simplified case, to make visualization easier. Suppose the black box is deciding whether to approve mortgages, and it only has three inputs: income, credit score, and mortgage amount.
You can visualize the inputs as points in a 3-dimensional space, each point representing a unique combination of income, credit score, and mortgage amount.
Now imagine in that space we place green dots at all the points that represent inputs that result in approval of the mortgage application. Do nothing at the points for inputs that represent rejected inputs.
Need to give someone an explanation of their rejection? Find their point in that space, and then find points on the green region that are nearest their point, and tell them what would need to change to move their point to those green points.
I'd suggest, in this 3-dimensional space, to look at four green region points. (1) the nearest green point reached moving from their point parallel to the income axis, (2) same but using credit score axis, (3) same but using mortgage amount axis, and (4) the nearest green point.
In the context of a decision not made by an algorithm, would you find acceptable an explanation like "many repeated studies from well-known and reputed institutions and universities have concluded that X leads to Y, and therefore as a way of providing Y to our customers, we have chosen to do X"?
In case your answer is yes, would it be acceptable to use the track record of a particular network and similarly trained networks, and why or why not?
Rather like the old joke about a banker who will lend you an umbrella while it's sunny but wants it back when it's raining, it seems that the health insurance business thrives on trying to avoid as far as possible giving healthcare to people once they're sick.
I haven't heard the original joke in full, but the way you wrote it, it very much applies - umbrellas are somewhat useful when it's sunny too, so you get some benefit from lending one. Similarly, health insurance isn't useless, but it definitely tries to not be there when needed most.
>Definition of a banker: A man who loans you an umbrella when the sun is shining, asks for it back at the first sprinkle of rain, and doesn’t own the umbrella in the first place.
Tip: If you are very casual smoker and see a form asking you "Have you ever smoked?" or "Do you smoke any tobacco?" then say no. If you say yes then you will see it getting reported to your health insurance and even life insurance and your premiums will go up as if you were chain smoker.
I'm not sure that's a good tip, presumably if you get lung cancer they can still detect if you are or have been a casual smoker for a non-trivial amount of time. In addition, you might put 'No' on signup and be accepted, but a lot of due diligence (uncovering your real history) is done at the time of claim i.e. they contact any past doctors and all surrounding hospitals for medical reports related to your past. Admittedly there is a lot more leeway for insurance companies in Asia to pursue their denials any way they see fit since we don't have a first world grasp of privacy or anything remotely like GDPR.
Hope that's obvious. If you are going in for something related to smoking always talk to your doctor. I'm talking about simple stuff where they give you large forms on what you have. They form end up at insurance companies and your rates would reflect that.
Today, maybe. Btw, a side line to the point as do qualify your defintion, other peoples'' definition of casual and very casual vary a lot, I know people that if a pack lasts more than one day they consider that casual.
I have seen a few definitions of casual smoking, I meant (one cigarette per week). I have seen not smoking every day, averaging less than one cigarette per day used.
"Pack year" is often used as a risk assessment: https://en.wikipedia.org/wiki/Pack-year 1 pack year in a lifetime is not safe, but it is a vastly lower risk profile than a pack+ per day smoker.
From the article: This allows the insurance companies to keep insurance costs down for both the insured and the insurers.
No! Mathematically the insurance business works the same as long as the average premiums cover the average costs for the pool. Identifying higher risk individuals and charging them more does nothing to lower the average cost as long as they continue to buy insurance. It only spreads the costs around differently.
That is, unless the idea is to make the cost so high for risky patients as to force them not to participate in the pool.
Which is I guess what they are saying. It does lower average premiums in the pool the only way mathematically possible, by forcing at-risk or sick patients out of coverage.
This raises average costs for the original pool (from before some patients were priced out) because those patients forced out will show up later at emergency rooms with expensive problems and no coverage. But that is no longer the insurance company's problem, so who cares.
Well at a society level there is no forcing people out other than death. So unless we can get enough sick people to die quickly and cheaply this system just forces higher costs on everyone.
Or maybe we could just cover everyone in one national pool without keeping a secret database of everyone's health history to use against them and gouge them more?
Say insurers know nothing about customers, just an anonymous number. In this world they would have to charge everyone the same, the average cost of insuring a person. Everyone pays the same premium, and has all their medical bills paid.
In a parallel world, insurers have perfect information. They know your medical future, not just the history. In this world, insurers will price your premium at exactly the cost of paying your medical costs. If you won't have to go to the doctor this year (this is a world where insurers know the future), you won't be charged anything. If you're going to be in a car accident, you will be charged for surgery and rehab.
In world no. 2 where insurers have maximum knowledge, insurance basically doesn't exist. Insurance is exactly the same as just paying your bills directly. In world 1, insurance works fine, but a 19 y/o pays a little more than they do today and sick or elderly people pay less than in the real world.
As data/information becomes more ubiquitous, we travel down this continuum.
Yup. I wish we could travel up this continuum, but then the real problem shows up: you need to figure out a way to stop health service providers (and their providers) from sucking your general insurance budget dry. Otherwise, you'll find that an increasing amount of people can't afford their payment.
The solution to this problem is single payer. Where the insurer (government) now has negotiating power on prices because they’re the only game in town.
That is one solution to one problem. It has the side effect of overly simplifying American systems of governing and labor, and would cause new side effects or amplify existing ones.
The US government is essentially the only game in town for military hardware and contracting, but I don't think anyone would hold up that procurement system as a model of efficiency.
In the world of health care, many developed countries have single-payer healthcare and the US pays twice as much for everything as all of them. If goverment-run health care is inefficient, why is our "free-er" system half as efficient?
I am all for single payer (Medicare for all for example) but there are plenty of ways the US system could improve. First would be to mandate pricing transparency so we actually have some data on the table for a rational discussion.
In the long run its the same funding model as police and public schools, so health care will mostly be funded by property tax (or crazy local alternatives for those without prop tax), and the same local people who balance the tradeoffs of protecting from criminals vs protecting from fire will now be balancing education vs health care.
Virtually all health care is local and monopoly and time sensitive, just like policing or kindergarten.
I live six blocks from supposedly the best neonatal intensive care unit in the region; Only people who live nearby can benefit and I should pay more for it. Much like I pay a lot of prop tax for the local police to keep crime rates to low range of the suburban experience. Much like education I'm sure the feds and state will kick in some bucks but most of the money will come from property tax (much as most of the benefits to me come from living nearby)
I believe that funding based on local property tax is a uniquely American proposition.
In other countries, it is considered distasteful that people in poor neighborhoods get less funding for their schools and police force. And especially for schools, the country at large reaps the benefit from people going to a good school, much moreso than it is a benefit to parents that their children go to a good school.
Yes foreigners think majority funding means 100% funded, but my local K12 gets $80M from prop tax, $50M from the state acting as middleman (its a proxy for income tax funding the schools), and $3M from the feds.
With the rise of identity politics in the USA, there's little motivation to help the kids of people who hate you and your kids, its a natural outcome of a multicultural nation.
Poor communities of all races have been underserved since long before whenever you imagine "identity politics" started to "rise" in the USA. Was "identity politics" not enshrined in the US Constitution, which protected racial slavery while giving more-slaveowning states extra voting power over less-slaveowning states
in world 1, 19 year olds pay a lot more than they do now, unless we are meant to assume that world 1 significantly increases the efficiency of the whole healthcare system. one of the stipulations of obamacare is that premiums for the elderly can be no more than three times the premiums paid by healthy young people. this implies that the natural ratio is greater than 3:1.
Assuming the future is known, there is no risk and therefor need to take an insurance. It's better to keep the money and remove the insurance admin overhead.
Call me cynical but in practice I think that kind of asymmetric information is only going to serve the insurer to rake in bigger benefits.
The pricing game is between the insurance companies and the providers; the end recipients of services have no involvement in the decision making process.
Its kind of like cattle, veterinarians, and ranchers. The cattle don't have much of a say in the finances or what treatments they get. The ranchers want them to feel good to make them easier to control, but ...
Yes true. I was just trying to demonstrate a point, and it's easier with absolutes.
The real world scenario is that a combination of better startistics/ML, and access to a lot more data moves us increasingly closer to that world where medical insurance doesn't make any sense. Insurance works for unknown/unknowable situations.
Spontaneous combustion insurance is cheap because it could happen to anyone.
I disagree. The “the lucky should share in the costs of the unlucky” aspect of insurance still is there; the difference is that, in a perfect information world with self-serving individuals, the need still is there for the unlucky, but the lucky won’t be willing to take part.
I'd say we could have world #1, but the C level execs would have to make salaries that are for more in line with the kind of pay that the rank & file make. And the companies would no longer be publicly-traded.
I'm not for government takeover of health insurance, but am very much in favor of bringing the rent seekers back down to earth to struggle in the muck with the rest of us.
Hilarious! I work for a major publicly-traded health ins. co. Our CEO is one of the highest paid people in the state, so is the 2nd in command, and 3rd, and a few more down from there. It's insane how much money is eaten up in ways that do nothing at all to insure individuals, but they make big talkie talk about all of it.
I would say that in my organization, for every 10 people that work here, only 2 do actual work.
It's not about forcing sick people to die, it's about an inefficient corporation that soaks up billions to return very little in actual value.
BTW, I have always prided myself with my ethics in the sense that I would never work in the "defense" sector, etc. I am thinking that big healthcare is a breach of this same type of ethic. I am at a crossroads in my career as a developer, and I really want to just get away from it all, but I'm 45 with two kids and I need at least another 5 years of work before I can downsize and escape the corporate grind for good. It sucks, though, I won't spritz it up. We should just have a revolution in the US.
Insurers with more data can create incentives to stay healthy. E.g. insurer could tell you "if you lose 10 pounds your insurance is going to decrease by X".
as long as the average premiums cover the average costs for the pool
Not even. In P&C insurance at least, the premiums are invested. Even if you operate at an actuarial loss (i.e. you pay out more than you receive in revenue), you can still make a profit on the investment income.
The article doesn't address whether an individual can request a full report of their own records and those of their dependents from this MIB. I can get a copy of my credit report for free, same should apply here.
thanks for sharing excellent post and Compare Rates for Business Loans in Chandigarh and Get Best of Deals. Get Best of EMI per Lacs on Business Loans in Chandigarh. Compare and choose from Top Business Loan providers in Chandigarh read more http://www.retailloans.in/business-loan-in-chandigarh
In the UK, health insurance companies buy data from supermarkets like Tesco (gathered from loyalty cards) to understand your lifestyle and potential risk by viewing what you eat and drink over a period of years.
55 comments
[ 3.6 ms ] story [ 127 ms ] threadThe article in this post, however, is lacking an explanation of how MIB gathers the medical records. It is therefore unclear if the users gave their consent to share the information even if it was through vague fine print. On the other hand the story would be very different if the data was gathered and shared without consent and I might wonder if a legal case can be made here.
[0] https://stallman.org/ancestry.html
So basically, if you don’t want your information in MIB don’t apply for insurance that requires a medical history.
Various third parties have real-time access to your hospital admissions and prescriptions, in some cases before your insurer.
As an example, when my wife was admitted to the hospital at 8 weeks for a miscarriage that almost killed her, the level of detail provided to data brokers allowed Enfamil to Fedex a big sample box of formula to us on what would have been the due date of the baby.
No consent was given or needed.
[0] https://arxiv.org/abs/1606.08813
[1] https://news.ycombinator.com/item?id=12048223
I remember when Alphago was beating the go master, we couldn't really tell why it made certain moves.
But it certain makes sense to require an explanation for things that affect us in our lives.
If you're making algos for that directly affect another human's basic needs, I think it's really important to be able to back up why the algo did what it did. I feel like you're probably opening yourself up to massive liability (regardless of region your operating in) if you can't explain to a court why your system did X to a person that resulted in them suing your company.
Which is the exact result people want. We want explanations that are valid and understandable in relatively plain language, using postulate logic.
Let's take simplified case, to make visualization easier. Suppose the black box is deciding whether to approve mortgages, and it only has three inputs: income, credit score, and mortgage amount.
You can visualize the inputs as points in a 3-dimensional space, each point representing a unique combination of income, credit score, and mortgage amount.
Now imagine in that space we place green dots at all the points that represent inputs that result in approval of the mortgage application. Do nothing at the points for inputs that represent rejected inputs.
Need to give someone an explanation of their rejection? Find their point in that space, and then find points on the green region that are nearest their point, and tell them what would need to change to move their point to those green points.
I'd suggest, in this 3-dimensional space, to look at four green region points. (1) the nearest green point reached moving from their point parallel to the income axis, (2) same but using credit score axis, (3) same but using mortgage amount axis, and (4) the nearest green point.
In case your answer is yes, would it be acceptable to use the track record of a particular network and similarly trained networks, and why or why not?
https://quoteinvestigator.com/2011/04/07/banker-umbrella/
I personally like the 1949 version:
>Definition of a banker: A man who loans you an umbrella when the sun is shining, asks for it back at the first sprinkle of rain, and doesn’t own the umbrella in the first place.
That's not just health insurance. All insurance wants to take your money and then avoid paying in any way possible.
"Pack year" is often used as a risk assessment: https://en.wikipedia.org/wiki/Pack-year 1 pack year in a lifetime is not safe, but it is a vastly lower risk profile than a pack+ per day smoker.
No! Mathematically the insurance business works the same as long as the average premiums cover the average costs for the pool. Identifying higher risk individuals and charging them more does nothing to lower the average cost as long as they continue to buy insurance. It only spreads the costs around differently.
That is, unless the idea is to make the cost so high for risky patients as to force them not to participate in the pool.
Which is I guess what they are saying. It does lower average premiums in the pool the only way mathematically possible, by forcing at-risk or sick patients out of coverage.
This raises average costs for the original pool (from before some patients were priced out) because those patients forced out will show up later at emergency rooms with expensive problems and no coverage. But that is no longer the insurance company's problem, so who cares.
Well at a society level there is no forcing people out other than death. So unless we can get enough sick people to die quickly and cheaply this system just forces higher costs on everyone.
Or maybe we could just cover everyone in one national pool without keeping a secret database of everyone's health history to use against them and gouge them more?
Say insurers know nothing about customers, just an anonymous number. In this world they would have to charge everyone the same, the average cost of insuring a person. Everyone pays the same premium, and has all their medical bills paid.
In a parallel world, insurers have perfect information. They know your medical future, not just the history. In this world, insurers will price your premium at exactly the cost of paying your medical costs. If you won't have to go to the doctor this year (this is a world where insurers know the future), you won't be charged anything. If you're going to be in a car accident, you will be charged for surgery and rehab.
In world no. 2 where insurers have maximum knowledge, insurance basically doesn't exist. Insurance is exactly the same as just paying your bills directly. In world 1, insurance works fine, but a 19 y/o pays a little more than they do today and sick or elderly people pay less than in the real world.
As data/information becomes more ubiquitous, we travel down this continuum.
Virtually all health care is local and monopoly and time sensitive, just like policing or kindergarten.
I live six blocks from supposedly the best neonatal intensive care unit in the region; Only people who live nearby can benefit and I should pay more for it. Much like I pay a lot of prop tax for the local police to keep crime rates to low range of the suburban experience. Much like education I'm sure the feds and state will kick in some bucks but most of the money will come from property tax (much as most of the benefits to me come from living nearby)
In other countries, it is considered distasteful that people in poor neighborhoods get less funding for their schools and police force. And especially for schools, the country at large reaps the benefit from people going to a good school, much moreso than it is a benefit to parents that their children go to a good school.
With the rise of identity politics in the USA, there's little motivation to help the kids of people who hate you and your kids, its a natural outcome of a multicultural nation.
Poor communities of all races have been underserved since long before whenever you imagine "identity politics" started to "rise" in the USA. Was "identity politics" not enshrined in the US Constitution, which protected racial slavery while giving more-slaveowning states extra voting power over less-slaveowning states
https://blackpast.org/aah/three-fifths-clause-united-states-...
https://en.wikipedia.org/wiki/Brown_v._Board_of_Education
Call me cynical but in practice I think that kind of asymmetric information is only going to serve the insurer to rake in bigger benefits.
The pricing game is between the insurance companies and the providers; the end recipients of services have no involvement in the decision making process.
Its kind of like cattle, veterinarians, and ranchers. The cattle don't have much of a say in the finances or what treatments they get. The ranchers want them to feel good to make them easier to control, but ...
The real world scenario is that a combination of better startistics/ML, and access to a lot more data moves us increasingly closer to that world where medical insurance doesn't make any sense. Insurance works for unknown/unknowable situations.
Spontaneous combustion insurance is cheap because it could happen to anyone.
I disagree. The “the lucky should share in the costs of the unlucky” aspect of insurance still is there; the difference is that, in a perfect information world with self-serving individuals, the need still is there for the unlucky, but the lucky won’t be willing to take part.
In that system everyone who pays the insurance would be better off putting the money aside as they wouldn't have to pay for the admin overhead.
Even if the catastrophic event happens early in the timeline it would probably be better to take out a loan than to pay the insurance.
I'm not for government takeover of health insurance, but am very much in favor of bringing the rent seekers back down to earth to struggle in the muck with the rest of us.
I would say that in my organization, for every 10 people that work here, only 2 do actual work.
It's not about forcing sick people to die, it's about an inefficient corporation that soaks up billions to return very little in actual value.
BTW, I have always prided myself with my ethics in the sense that I would never work in the "defense" sector, etc. I am thinking that big healthcare is a breach of this same type of ethic. I am at a crossroads in my career as a developer, and I really want to just get away from it all, but I'm 45 with two kids and I need at least another 5 years of work before I can downsize and escape the corporate grind for good. It sucks, though, I won't spritz it up. We should just have a revolution in the US.
Not even. In P&C insurance at least, the premiums are invested. Even if you operate at an actuarial loss (i.e. you pay out more than you receive in revenue), you can still make a profit on the investment income.
Hmmm ... might be able to get a copy here: https://www.mib.com/request_your_record.html