joshgel

Also, lots of other quality of life and expense factors to consider: schools, dinning/produce, jobs in your field.

Ultimately though, doesn’t the market just work this out? The areas where the most people want to live are the most expensive and visa versa. The trick is figuring out what matters to you and what doesn’t and then finding the cheapest place that offers what you want. With the obvious caveat that for many many people, being close to friends and family is a major factor in what they want.

Lots of people in the medical/scientific community expected this result. The best explanation I have heard is that the vaccine companies planned these sub-optimal trials with 2-3 weeks between shots, because it meant getting the vaccines to market much more quickly. Instead of having vaccines available in December, they might not have been available until February.

So far, the evidence is pretty compelling that the vaccines have continued efficacy 6-7 months post-administration. It's obviously impossible to know much beyond a couple months however.

That didn’t last long

For what its worth, the A16Z boss talk podcast episode/clubhouse this week had everyone (incl Databricks CEO) basically agree with this model. Most people in the office 2-3 days/week. Some full remote.

Seems like that is the way the valley is headed.

In the paper associated with 1 it says:

The FilmArray RP test utilization rate (TUR) metric is defined as the non-normalized number of RP patient test results generated each week across the Trend sites (computed as a centered 3-week moving average). To calculate the pathogen detection rate (as displayed in Figure 2 [second data view] and on the Trend website), we compute the rate for each organism at each institution as a centered 3-week moving average. To adjust for the capacity differences between sites, a national aggregate is calculated as the unweighted average of individual site rates. Only data from sites contributing more than 30 tests per week is included to avoid noise from small numbers of tests. Because the calculation of pathogen detection rate includes results from patients with multiple detections, the detection rate for all organisms can, in theory, add up to greater than one. In practice, this does not occur.

These are rates. This does not account for testing frequency. If they only tested 1 person in the United States tomorrow, and they had rhino/entero, the graph would show 100% (obviously it wouldn't and they won't only test 1 person). That means that if testing decreased, because for example there were less patients with symptomatic viral infections, but the % positive stayed the same, the chart would not appear to change, even though there were fewer cases...

paper: https://publichealth.jmir.org/2018/3/e59/

2 was published in Feb 2020, so not sure how helpful that is...

1 is not normalized data. It does not account for testing frequency. These percentages are of tests using BioFire. My hospital, for example, does not use BioFire to test for SARS-CoV-2 and so we end up running BioFire much, much less frequently than during a normal year.

Anecdotally non-COVID respiratory infections in hospitalized patients have decreased dramatically. Working on trying to demonstrate this with data.

Agree the tax calculations are a little bit off. Also, only lets you take the individual standard deduction (12k vs 24k for married filing jointly)

Similarly, the ability to share with another account, so that my spouse can also poke around and create models based on our base case.

This is possible, but I work in medical research, and this pause is pretty standard practice. When there is a potential problem, you pause, gather more data, come up with a mitigation plan (if needed) and submit all this to the ethics authority (usually IRB). Its rote standard practice. And i think much more likely that they are following standard practice than there are hidden issues that haven't otherwise been revealed publicly yet.

That said, this is not a usual situation at all, and I think it's absurd, not least because of lives potentially saved during this pause, but because of the carry on effects with vaccine hesitancy, etc.

Also, if you phrase the question “should everyone follow the same rules?” in the right way, everyone is going to say “of course!”

I saw that too in, for example, Lower New York Harbor.

Do all Western democracies guarantee Freedom of Speech like in the US? I seem to remember reading this is actually a US-centric view that isn’t universally shared. Eg nazi slogans in present day Germany.

Do we actually need to guarantee universal Freedom of Speech to have a democracy?

This is the beauty of randomization and blinding. Because both the participants and the doctors (the ‘double’ in double blind) don’t know who got the vaccine or placebo, anything you tell them to do is told equally to both sides. You can’t alter the trajectory of one group over the other because you don’t know the groups.

Right, I'm not sure I should get dinged for letting my daughter listen to Baby Shark or Taylor Swift on repeat, but I guess that's life...

I recently wrote a short example based gist for a specific health care use case (evaluating FHIR data) using jq.

https://gist.github.com/joshgel/12082d23a75feaab5d405db31981...

Is this true? My understanding was that in fine tuning, you’d only re train some of the layers. And even if you re train all the layers, the starting point for the layers is not random. If it really was all forgotten then fine tuning would not be orders of magnitude faster...

You do not have to be in a high risk group. I've seen people not in high risk groups die. Your odds may be lower, but no one is immune. (I guess except those who survived it or got the vaccine)

I'm firmly in the camp that exercise is extremely beneficial. Likely one of the most beneficial things one can do (with not smoking and eating a diet high in fruits/veggies low in processed carbs).

We (society/physicians/etc) need tools to get people to exercise. This is more a study of how effective one of those tools might be (ie supervised exercise vs recommendations). If supervised exercise is no better than recommendations, I find that pretty disappointing. Because as we all know, recommendations don't work very well, or at least most people don't adhere to them.

Wow what a disappointing result. Perhaps underpowered to detect what may only be a small change.

Also perhaps some selection bias in that the participants that sign up for an Exercise Study may already be pretty fit. And we see some of that in the baseline characteristics.

This was a retrospective study. Meaning the 'general population' they drew from were people with vitamin D tests during the period (Jan-Mar 2020) already drawn for some reason.

Not sure that this is true. It might be. But more likely being active enough such that you get the 15 minutes of sunlight daily needed to maintain normal vitamin D levels is better advice.

Even bigger issue than base rate is that this is a retrospective study, so I suspect heavy measurement bias. Patients that are likely to have a Vitamin D level checked are different than hospitalized patients. Vitamin D levels are not checked routinely in hospitalized patients. I know this because I'm a hospitalist, but also because during the height of the local pandemic this 900+ bed hospital only had 216 patients with a vitamin D level checked. Presumably at least 1000 COVID patients moved through that hospital during this time, more likely several thousand.

Previous discussion here: https://news.ycombinator.com/item?id=24622280

I've searched extensively for any studies that describe the 3 month follow up and found exactly one, recently published. I would love if you could identify more.

Results aren't really as described here. Some patients recover, but about 1/3 had persistent lung findings and many had decreased lung function (as measured by DLCO).

[0] https://www.thelancet.com/journals/eclinm/article/PIIS2589-5...

With tens of thousands of publications on COVID, its absolutely amazing how few there are on the follow up of these patients. (ie what comes next after infection). There is so little data we are flying blind (again).

That said, there is evidence that the other coronaviruses that caused epidemics (SARS and MERS) can cause clinically meaningful myocarditis and heart failure as a result. In vitro and autopsy evidence suggests that SARS-CoV-2 directly infects myocardial cells. So it's plausible that there will be some cardiac sequelae of COVID, perhaps in many patients or maybe just in those at high risk. We just don't know. Assuming that myocarditis will be a major problem because 2 uncontrolled studies said so is as silly as assuming hydroxychloroquine works because of similar strength of evidence. But dismissing it out of hand (which I'm not suggesting GP is doing) is not smart either. We should be appropriately skeptical, but need to investigate further.

Certainly hard to measure, but Figure 2A certainly looks like an adherence chart.

Regardless, in RCTs we usually consider a 'intention to treat' analysis gold-standard. That means that since you are assigning an intervention in the real world, even if someone drops out, you still consider them part of the intervention group, because the same thing will happen in the real world. You can also do a 'per protocol' analysis, which attempts to measure the impact of the intervention to those that adhered, but not as important...

as i said lower down: "ehhh, really depends on what effect size you are looking for. presumably you would want to see a 10% effect size or more. (detecting that intermittent fasting caused 2 lbs more weight loss than CMT would be statistically valid, but not clinically that important)"

power depends not just on n but also on the expected difference between the means (and the standard deviation of the mean)

ehhh, really depends on what effect size you are looking for. presumably you would want to see a 10% effect size or more. (detecting that intermittent fasting caused 2 lbs more weight loss than CMT would be statistically valid, but not clinically that important)

This. The EMR does go down from time to time. We still have paper order pads around so that things can get done. Its annoying AF. You have to fax things to radiology to get a scan. You have to walk to pharmacy to get meds. But, the system worked before computers (which wasn't that long ago in many hospitals) and it works without them today.

That said , the first time I had to use paper I felt like I had no idea what I was doing, because a lot of the EMR defaults aren't there to prompt you. So you have to specify a 'start time' for a medication, but in the EMR now is the default, so I didn't think about it as anything mandatory.

Don't we just end up with a balkanized internet? A Euro-net, US-net, China-net?

I guess Canada goes in with Euro-net?