Curing cancer statistically via mammography

One of the papers that we studied during our Harvard Medical School “big data” course in February was “National Expenditure For False-Positive Mammograms And Breast Cancer Overdiagnoses Estimated At $4 Billion A Year” (Health Affairs 34:4, 2015). The researchers used a data set of 4 billion insurance claims to see what was going on in the U.S. population. We learned that screening mammograms are not helpful compared to waiting for a lump to show up. There are a lot of things that look bad on a mammogram that aren’t, in fact, bad.

Americans fell in love with mammograms:

Why do we love them so much? It turns out that the five-year survival rates for breast cancer were improved after women en masse got put through the mammography industry. Why would anyone want to stop doing something that improved five-year survival rates?

It turned out that the statistical cure for breast cancer because of mammography was due to the fact that women who did not have cancer were being treated for cancer. They hadn’t been killed by cancer five years later because… they never had cancer to begin with.

So we wrapped ourselves around the axle with data that we weren’t smart enough to comprehend.

(Separately, we learned during this medical school class that it takes approximately 17 years for an identified “best practice” to be adopted by physicians nationwide. Thus we can expect Americans to back off on their love for mammography perhaps in 2032.)

Related:

Full post, including comments

Reconfigure our National Parks as urban environments?

School is almost out, so it is time to think about retreating to the peace and quiet of our National Parks, right?

Check out these photos from my October 2017 trip to Zion National Park. The parking lot at the park entrance is generally full so you park a few miles away and wait for a shuttle bus, packed to standing-room-only. Then you wait 30-45 minutes to get on a second shuttle bus, within the park, and that too will be standing-room-only. The visitor center features a 20-minute line for “information” next to a sign talking about “wilderness” and how it “has outstanding opportunities for solitude”.

Zion National Park was established in 1919 and I don’t think that any new trails have been built since then. Consequently, once you do get off the shuttle bus, unless you’re planning to do a 15-mile hike, you’ll likely be on a trail that is more crowded than the average Manhattan sidewalk (not more crowded than Times Square, of course, but far more crowded than the sidewalks away from Midtown). See my photos for the typical crowds on a trail in a shoulder season. The monopoly vendors of crummy food enjoyed long lines of customers, the flip side of which was that it was much more time-consuming to purchase food than it would be in Manhattan, even during the Midtown lunch rush.

Currently access to the park is rationed by ability to stand in lines, ability to stand on a packed bus, and tolerance for walking on a jammed trail. (We went to a convention in Las Vegas afterward, attended by nearly 30,000 people, and it seemed far less crowded than the National Park.)

Americans don’t seem to be willing to ration access to this scarce resource by price. See, for example, “National Park Service Reconsiders Steep Fee Increase After Backlash” (nytimes):

A Trump administration proposal to steeply increase entrance fees to the most popular national parks landed with a thud when it was presented in November, and park officials say they are now reconsidering it.

The proposal, which would apply during the peak visitor season to 17 parks including the Grand Canyon, Yellowstone and Yosemite, called for a $70 fee for noncommercial vehicles, up from $30.

The fee increase seems to be sold as a way of increasing the funds available for maintenance, not as a way to reduce crowds.

Since we aren’t going to reduce the crowds by imposing a higher price and we’re going to continue to grow U.S. population through immigration and incentives to Americans to have more kids, our National Parks are going to end up being more like tourist attractions in Japan or China. Once we accept that we’re going to have Shanghai-style densities, why not reconfigure the parks to be more like Shanghai? Build a denser network of trails where possible. Run robot-driven buses every minute (like the Moscow subway!). Put in high-rise hotels and McDonald’s restaurants that are capable of handling a high volume of customers efficiently. Since we have accepted that our National Parks should be urban environments, why not run them as competent urban environments instead of as pretend wilderness with hour-long waits for the standing-room-only bus and 30-minute waits for a frozen burger?

Full post, including comments

Headline hysteria versus boring statistics: Flint water system

“The Children of Flint Were Not ‘Poisoned’” is interesting because it comes from the same nytimes that maintains a “Flint Water Crisis” section.

What are the boring statistics?

Right now in Michigan, 8.8 percent of children in Detroit, 8.1 percent of children in Grand Rapids and an astounding 14 percent of children in Highland Park surpass the C.D.C. reference level. Flint is at 2.4 percent. A comprehensive analysis of blood lead levels across the United States reveals at least eight states with blood lead levels higher than Flint’s were during the water switch.

Flint did hit 3.7 percent at the height of the “crisis,” but never reached anywhere near the level of Detroit, which is not in a “crisis”. But perhaps the measured increase did not correspond to an underlying change?

Moving from evaluating percentages to examining actual blood lead levels in children, we found that levels did increase after the water switched over in 2014, but only by a modest 0.11 micrograms per deciliter. A similar increase of 0.12 micrograms per deciliter occurred randomly in 2010-11. It is not possible, statistically speaking, to distinguish the increase that occurred at the height of the contamination crisis from other random variations over the previous decade.

Obviously any amount of lead is bad, but given that other cities have worse issues with lead and these data are readily obtainable, why the focus on Flint?

Related:

  • see this mini-site about our suburban neighbors’ plan to spend $100 million on a new school because they think there is lead in the drinking water in the old school; people cannot be convinced by negative lab tests nor can they be convinced to buy $75 filters that take out 99 percent of any lead (so these environmentalists have been trucking in jugs of water every week for the last 10+ years, but that isn’t Green enough so it is time to push the 1994-built/renovated school into a landfill)
Full post, including comments

Can we trust Customs and Border Protection to screen refugees and asylum cases given what they’ve done with their eAPIS web site?

Americans expect the Department of Homeland Security to do a complex task reliably, i.e., figure out if someone who appears at a U.S. border crossing and asks for asylum (or who crosses illegally and is arrested and then claims to be a victim of persecution back home) is telling the truth and is therefore entitled to the usual U.S. welfare package (i.e., a lifetime of free housing, free health care, free food, and free smartphone).

It is impossible for citizens to know if this task is being done competently. Generally we don’t speak the language of the migrants. We don’t have any way to verify if their stories are true (see this story about a mother of four young children who says that she is being persecuted by the Honduran government; how would we ever determine the truth or falsehood of this statement?).

But most of us know what a competently-run web site looks like. About 13 years ago the U.S. government decided to impose on private aircraft operators the requirement to pre-submit passport-type details for everyone departing or arriving in the U.S. This is the “eAPIS” system at eapis.cbp.dhs.gov. We know that it has a database back-end because it can remember pilot details and information from recently filed manifests, but it is impossible for a family to enter all of the non-pilots persistently. This has led to various subscription services ($250/year for a popular example at fltplan.com) where someone else will keep a database and send over a completed manifest to CBP via XML. You might think that in 13 years the programmers at or working for CBP would have added the most-requested features, such as the ability for each user to save details for a few friends or family members, but this has not happened. At a minimum, this would likely reduce transcription errors (if a passport number is entered once it is more likely to be correct and consistent than if it must be entered 25 times).

The argument, I guess, is “well, this CBP agency is terrible at running a web service, but they’re great at everything else they do.” But usually when an enterprise is good at one thing they are pretty good at everything and when they’re bad at one thing it is usually because management has low standards for pretty much everything that the enterprise does.

Can we infer from their inability to run a decent 1995-style web service that CBP is never going to be able to screen refugees and asylum-seekers?

Full post, including comments

Sex on the job, American versus Russian editions

“Maria Butina, Suspected Secret Agent, Used Sex in Covert Plan, Prosecutors Say” (nytimes) describes how a 29-year-old student pilot in a Piper Warrior (see the video on the page at 2:11) managed to subvert the $4.4 trillion/year Federal government:

Apparently hoping for a work visa that would grant her a longer stay, she offered one American sex in exchange for a job. She moved in with a Republican political operative nearly twice her age, describing him as her boyfriend. But she privately expressed “disdain” for him and had him do her homework, prosecutors said.

In other words, a young person was having sex with an older person who had something to offer in terms of career advancement.

When this happens between two Americans, the young person is described as a helpless victim, suffering from what might be a year of forced sexual encounters, utterly trapped and with no thought of what was being obtained in exchange. When the young person is Russian, however, the same newspaper describes the young person as a cunning aggressor, in control of the situation.

Also in anti-Russian hysteria:

  • “NRA ADMITS ACCEPTING MONEY FROM 23 RUSSIA-LINKED DONORS” (Newsweek), in which we learn that it is front-page news for Americans that, over the past 3 years, $2,500 was received, mostly in subscription or membership fees, from 23 people ($100 each!) who “may include U.S. citizens living in Russia.”
Full post, including comments

Harvard Humanitarian and Affordable Housing

Harvard makes the Daily Mail: “‘Are you in one of the affordable units?’ Director of the Harvard Humanitarian Initiative is accused of being a racist ‘Sidewalk Susie’ after slamming a mom and her biracial daughter, 3, for playing outside”

Cambridge has taken to calling its “steal from the (rich) developers and give to the poor” program “inclusionary housing”:

The Inclusionary Housing Ordinance applies to new residential developments or buildings converted to residential use which create 10 or more new housing units or over 10,000 square feet of residential space. The Ordinance requires that 20% of the residential floor area in a building be devoted to affordable units.

There are apparently roughly 1,100 units currently (source). This is in addition to the Cambridge Housing Authority, which provides 5,500 free and/or subsidized housing units. In a city of 110,000, however, in which most people with middle-class jobs have been forced to exile themselves to boring suburbs and unendurable commutes, there is near-infinite demand for free housing. The waiting lists therefore can stretch to infinity unless an applicant has some sort of hook into the system. Thus 99.99 percent of Americans are excluded, as a practical matter, from these “inclusionary” houses.

Full post, including comments

Should California split up its government without splitting itself into multiple states?

“California Supreme Court Squashes Bid To Split State Into Three” (NPR):

California will be staying in one piece, at least for now, after the state’s supreme court ruled that a proposal to divide California into three cannot be placed on the ballot in November.

… the Planning and Conservation League (PCL), a nonprofit environmental group in California, filed a lawsuit to block the measure from getting to a vote.

The PCL says the proposal amounts to a change in the state constitution, which must be decided by the Legislature before it goes to voters, The Sacramento Bee reports.

“We opposed the measure because splitting the state up doesn’t solve any of our existing challenges, it just makes them worse,” Howard Penn, the executive director of PCL, told NPR via email. “We have worked hard for the past 50 plus years on statewide policies that help make California a healthy place to live. If the state gets split up into three parts, it would be like hitting the reset button and starting over.”

I’m not sure why the judges wouldn’t let this on the ballot. The other 49 states would have had to approve the measure, right? And that would never have happened? So why does anyone care?

What about an internal split, though, that would be less dramatic in name but have more impact in reality?

Right now someone in San Diego who thinks that schools should be funded more lavishly or run differently has to travel up to Sacremento and argue with imperial overlords. With a population of 40 million, far larger than most European countries, there is no reason for anyone in the capital to listen to a citizen unless that citizen is crazy rich or famous.

A Pattern Language (written by a UC Berkeley professor and friends!) suggests a country sized like Denmark, about 5 million, if individuals are to have any practical chance of influencing laws and decisions. The claim is that people will be happier in a country where they have a political voice.

Why couldn’t Californians vote to take apart their unaccountable state-wide bureaucracy one piece at a time? For example, they could vote to disband the California Department of Education (“We oversee the state’s diverse public school system, which is responsible for the education of more than six million children and young adults in more than 10,000 schools with 295,000 teachers. We are in charge of enforcing education law and regulations and continuing to reform and improve public school programs.”). Adjusted for demographics, California runs the worst-performing schools in the United States (nytimes). With nothing to lose, therefore, just say that whatever the statewide department was doing would now be done by counties (smaller ones could band together and cooperate if necessary) and that funding would be similarly devolved onto the counties (they would still have the Federal education bureaucrats to report to, of course, just like now!).

If decentralizing control of schools worked out well, voters could move on to the next government function. San Francisco could continue to run a winner-take-all family court system (the current law; see also how it could have worked for Ellen Pao) while San Diego adopted shared parenting and capped child support profits (see this chapter of Real World Divorce for the three basic types of custody dispute resolution systems in the U.S.) There could be marijuana in supermarkets up in Eureka and restrictions on the sale of this critically important medication down in San Diego. The speed limit on highways could be set by county. Maybe out in the desert it be over the current statewide 70 mph limit. Back in Silicon Valley the limit could be set to 25 mph to reflect (a) the reality of actual travel speeds during most of the day, and (b) the experimental nature of the self-driving cars that are sharing the road. (That’s another good thing to regulate locally: self-driving cars! They make a lot more sense in some places than others and how does a bureaucrat in Sacramento know?)

Eventually the only things left at the state level would be the National Guard and the soon-to-be-$1 trillion high-speed rail project. There is nothing in the U.S. Constitution, I think, that requires individual states to have central bureaucracies and uniform statewide laws for any given government function.

California readers: Would this make you and your neighbors happier? You’d still be in the same state as people who live 500 miles away, but you wouldn’t have to get their approval anymore.

Full post, including comments

How can bars offer discounts to women as a practical matter in the transgender age?

“A Fight for Men’s Rights, in California Courts: Ladies’ nights, career seminars and paternity fraud are all on the docket.” (nytimes):

California’s Unruh Civil Rights Act … outlaws discrimination against all people by any type of business establishment in the state, regardless of a person’s sex, race and other characteristics. Mr. Allison and his cohort would like to remind everyone that Unruh’s broad promise of “full and equal accommodations, advantages, facilities, privileges or services” extends to men.

The coalition members have become known around town as men who want to know when a bar or club is offering a discount on admissions or drinks only to women.

Maybe it is legal to have a different price for women as opposed to men, but how does it work in practice in the transgender and fluid gender age? If a patron walks in and says “I identify as a woman so I would like to pay the lower price” and the bar or club refuses, isn’t the bar or club then guilty of violating the rights of the transgendered? Even if discrimation against men is legal, isn’t it illegal in California to discriminate against those whose gender ID does not correspond to the sex printed on their birth certificate?

Related:

Full post, including comments

Product idea: automated scale-model Blue Angels and Thunderbirds air show planes

Here’s a product idea sparked by a summer of taking the kids to air shows: a box of six battery-powered airplanes that, given a soccer field of space, will recreate a Blue Angels or Thunderbirds air show. It seems as though the sensors and software of a DJI drone are already pretty close to being able to handle automated formation flight. Radio-controlled planes are capable of aerobatic maneuvers that are out of reach of even the fanciest full-sized planes. With another few years of costs coming down, would it be practical to put six jet-shaped planes in a box and sell it at Target for less than $200?

How awesome would it be for kids? Before or after a sacred suburban soccer game, the fully-charged set would (a) play the National Anthem (kneeling encouraged, but still optional), and (b) take off for a 15-minute formation aerobatics demonstration.

Full post, including comments

Historical aviation weather for accident or incident reconstruction

A lawyer contacted me the other day looking for an aviation expert witness. Ultimately I decided not to accept the engagement due to a lack of sufficient data for rendering an opinion, but I thought I’d write down what I learned about finding archived aviation weather, forecasts, and warnings.

The Plymouth State University Weather Center, run by good-hearted folks out of New Hampshire, offers archived radar and satellite images as well as surface charts.

Iowa State University offers archived pilot reports back to 2015 and National Weather Service text products, including aviation Center Weather Advisories back to 2008. NOAA itself, when not saving the whales, runs a “tape robotics” archive for NEXRAD and other data.

aviationwxchartsarchive.com offers a variety of charts and text products, including SIGMETs and SIGWX.

I couldn’t find anything for archived textual AIRMETs. If anyone knows of a source for these I would be grateful!

Service idea: A site where you can say “Give me a standard weather briefing as of Date X” Given that the NTSB already needs most of these data for its investigations, I’m surprised that the federal government doesn’t already run such a site (maybe the explanation is that the briefing web sites have been done by contractors and there is no point in a contractor innovating). The amount of data being generated per day is more or less fixed (old-school text format) and the cost of storage keeps falling, so this should become cheaper and easier to run over time.

Full post, including comments