pak

There are bound to be many more discovered as LLMs capable of doing so proliferate among those who don't report such things responsibly.

Keeping port 22 open puts you first in line for such exploits, while keeping it behind another layer (whether it's WireGuard or firewall tricks) would buy time, if not keep attackers away entirely. That seems useful, no?

Props to the OP for showing once again how lightheaded everybody gets while gently inhaling the GPT fumes…

https://github.com/powerpak/tqdm-ruby

(shameless plug and an invitation for pull requests)

Consider also the danger of having something dangling from your body that is powered by your arterial blood pressure (from a major artery, as the kidney is). A trip and fall could be instantly fatal.

https://www.cameronsworld.net/

I would love to see data on how common this phenomenon is in various populations of UTI patients (elderly, young, inpatient, outpatient, etc), given different prior exposures to antibiotics; for now it looks like 30 patients were assessed.

There are obviously many different causes of resistance previously identified, going all the way back to penicillinase enzymes inactivating penicillin. As often happens with lay summaries, this makes it sound a little too much like the cause of all antibiotic resistance has been found.

This post on ML in medicine is a pretty good overview of everything that has been going on recently and the nuances often lost in the current hype: https://lukeoakdenrayner.wordpress.com/2016/11/27/do-compute...

See this to understand why the bottleneck is residency positions, not how many US medical students there are: https://www.nytimes.com/2014/07/20/opinion/sunday/bottleneck...

>about 10% of people who come in to the cardiology clinic experiencing symptoms are diagnosed with an abnormal heart rhythm

OK, but I'd be more careful about staying apples to apples in your comparisons; your app is about asymptomatic AFib. So how many of those people going to the cardiology clinic had undiagnosed AFib; for how many of those would a new diagnosis of AFib have changed the plan of care; etc. Kind of like robbiep was saying, I would be interested in actual added value from the larger perspective.

Totally appreciate your point about perfect being the enemy of the good. The danger is that these semi-medical wearables currently straddle a strange zone between medical and consumer use. The inevitable marketing strategy is to co-opt the positive reputation of medical products while acknowledging none of the pitfalls of consumer products. Most of the screening methods you bring up are used by a doctor on symptomatic patients with a suggestive history, and only as a partial component of clinical judgement. The way Cardiogram seems to make the most money, on the other hand, is to sell the product to asymptomatic, casual users. (Furthermore, CHA2DS2-Vasc costs 30 seconds of talking or reading a medical record, not $700 in Apple products.) So you're inevitably running up against some doubts among physicians [0].

And finally, I agree that more machine learning practitioners should join medical research. I hope the field works to set more reasonable expectations, however, as in: ML will solve very specific subtasks in clinical reasoning (as in the diabetic retinopathy study [1]). Instead, the headlines usually ratchet that up to "AI will replace radiology/cardiology/$specialty in X years." That tends to hurt the people currently in the trenches, since their contribution in bringing about practical, incremental change is diminished. The top answer of this Quora thread [2] has a good discussion of the many dimensions of the problem.

[0]: https://twitter.com/Abraham_Jacob/status/860119573915287552

[1]: http://jamanetwork.com/journals/jama/fullarticle/2588763

[2]: https://www.quora.com/Why-is-machine-learning-not-more-widel...

1. Even at 98% sensitivity and 90% specificity [0], which I don't think would hold up with real world usage in casual, healthy users, if AFib has a prevalence of roughly 2-3% [1] then by a quick back of the envelope calculation a positive test result is still 5× more likely to be a false positive than a true positive. With those odds, I don't think many cardiologists are going to answer the phone. You'd still need an EKG to diagnose AFib.

2. There is huge variance among people's real world use of wearable sensors, and also among the quality of the sensors. (Imagine people that wear the watch looser, sweat more, have different skin, move it around a lot, etc.) You'd likely need to do an open, third-party validation study of the accuracy of the sensors in the Apple Watch before you can expect doctors to use the data. My understanding is that the Apple Watch sensors are actually pretty good compared to other wearable sensors, but I don't know of any rigorous study of that compares them to an EKG.

3. Obviously, this is only for AFib. AFib is a sweet corner case in terms of extrapolating from heart rate to arrhythmia, because it's a rapid & irregular rhythm that probably contains some subpatterns in beats that are hard for humans to appreciate. As others—including Cardiogram themselves [2]—have pointed out previously, many serious arrhythmias are not possible to detect with only an optical heart rate sensor.

[0]: https://blog.cardiogr.am/applying-artificial-intelligence-in...

[1]: https://www.ncbi.nlm.nih.gov/pubmed/24966695

[2]: https://blog.cardiogr.am/what-do-normal-and-abnormal-heart-r...

See the first five minutes of this video [3], which actually involves Apple's "Digital AV Multiport" adapter, to see the near-term experience of "dongle hell."

[1]: http://www.bourgedesign.com/

[2]: http://www.macrumors.com/2016/11/03/owc-debuts-13-port-thund...

[3]: https://vimeo.com/189525997

With only 2 ports, you need a ridiculous number of big dongles or hubs to get serious work done, and any pro user with >1 external HD, display, or gigabit wired is rightfully wondering what the hell Apple was thinking.

Before the MBP 2016 announcement: https://web.archive.org/web/20161028012214/http://www.apple....

And today: http://www.apple.com/shop/browse/home/specialdeals/mac/macbo...

Notice any differences?

Get them while they're still in stock. That's all I can say, it's what I just did. I cannot stand the new microwave-keypad feel of the new keyboard, and the lack of useful ports and price hike is just the extra slap in the face for me to wake up. This is probably going to be the last Mac laptop I buy.

Incidentally, I suppose the first 5 minutes of fighting with a three-way dongle and a poor connection, having to restart the talk, etc. is just a little preview of what every 2016 MBP owner will soon have the pleasure of experiencing themselves.

But the general point does hold that yes, a high level of math or CS training is advantageous for anyone moving into a career in the life sciences, since it appears that that's where much of the foreseeable growth (in careers and research funding) seems to be.

It's also been said by many that it's easier to learn some biology after training rigorously in CS/math, rather than the other way around. Dudley Herschbach (a Nobel-prize winning chemist) once said to me that his one piece of advice for young researchers would be simply, "Learn as much math as you can."

[1] https://en.wikipedia.org/wiki/Fidaxomicin

Sure, new methods of finding antibiotics are in the works, although the article you link has plenty of experts recommending caution about their potential. The bigger point is that in 2016 there is a looooooong road from antibiotic "candidate" to FDA-approved drug. That road involves decades of trials and costs billions of dollars per approved drug.

The larger problem is that there is little if any incentive for pharma companies to invest in antibiotics compared to traditional blockbuster drugs that are supposed to be taken chronically (and therefore have better ROI). It's the same reason little R&D goes into making new vaccines. It doesn't matter how many candidates are found if they can't make it to market in a timely fashion (the point of the CDC bar graph), and this is what the "slow catastrophe" really is. It is not that scientists will never figure out new ways to kill bacteria.

Secondly, there may be all kinds of other uses for glucose tests that one could research, but consumers running tests on themselves in an uncontrolled manner is not research. I would never say that no other uses will ever be discovered, but let's do that scientifically, please. My specific issue was with how diabetic glucose self-testing was used as rhetorical evidence that more blood tests help people, while failing to note that those tests are done to dose (potentially dangerous, fast-acting) medications, not to "keep tabs" on anybody's diabetes in a diagnostic sense, as was implied by the omission.

You say below that "people are coming around on glucose in the same way that we now understand that the cardio signal [...] are predictive of an enormous number of physiological and psychological phenomena." That's a lovely hypothesis, but please tell me who these people are, and please show me the evidence of the predictive value.

Until then, the Credentialed Professionals are perfectly justified in shrugging their shoulders at post-prandial glucose data from healthy patients (who, contrarily, will demand that needless and dangerous follow-up procedures are ordered for them), and the companies selling consumers these tests will not be helping anybody become healthier. I could go on, but this comment sums up the societal effects better than I could, even referencing your "ideal" of the ECG for screening. https://news.ycombinator.com/item?id=11694341

https://en.wikipedia.org/wiki/False_positive_rate

https://en.wikipedia.org/wiki/Positive_and_negative_predicti...

The latter depends on the prevalence of the test condition in the population, which is one of the major points of the OP.

A test can have a low false positive rate but still have a low positive predictive value if the test condition is sufficiently rare (as it is for most diseases). brianwawok was probably referring to tests with a low positive predictive value.

You're conflating a diagnostic test with a test that patients need to control dosing (of insulin). To make a diagnosis of diabetes, such frequent testing is not any more informative. Better tests, such as HbA1C, have been developed to indirectly measure blood glucose levels over a 3-month timescale, which is more appropriate for diagnosis.

I don't think there's any evidence yet that people being able to monitor their lipid levels while eating provides any useful medical information, unless you have some kind of (incredibly rare) inherited lipid metabolism deficiency.

comparing blood tests between a bipolar person's manic and depressive phases would be fascinating, but no one does it.

There in fact has been plenty of work on this, but in a research setting, where it belongs. See section 6 of http://www.ncbi.nlm.nih.gov/pubmed/27017833

it's the twenty-first century and the rate of iodine deficiency is 9%.

Micronutrient deficiencies are usually a result of dietary choices. This problem is more easily solved by encouraging everyone to take a daily multivitamin, which would be completely prophylactic, than by encouraging the same population to subscribe to series of blood tests that may or may not reveal the problem, and would require follow-up action. Again, think about it from a population health perspective.

What if there are some recommended follow-up treatments, but they are all expensive and risky, with the possibility of terrible complications?

This is how over-diagnosis actually leads to worse outcomes on a population scale. Medicine is often viewed as this big near-perfect algorithm where information is always enabling. In fact, too much information can be counterproductive for a patient and doctor.

See: https://www.statnews.com/2015/12/01/cesarean-section-childbi...