COVID-19: Exiting Lockdown and Geolocation

Pressure will mount around the end of this month (assuming we are past the peak death rate and virus spread is under control) for the US to exit lockdown. This needs to be done in a smart way, which includes:

1. Required use of facemasks
2. Cocooning of vulnerable populations
3. Contact tracing and forced isolation of cases, perhaps using geolocation technology

See related posts

COVID-19: Smart Technologies and Exit from Lockdown (Singapore)
COVID-19: CBA, CFR, Open Borders
COVID-19: Cocoon the vulnerable, save the economy?
COVID-19 Notes

WSJ: Western governments aiming to relax restrictions on movement are turning to unprecedented surveillance to track people infected with the new coronavirus and identify those with whom they have been in contact.

Governments in China, Singapore, Israel and South Korea that are already using such data credit the practice with helping slow the spread of the virus. The U.S. and European nations, which have often been more protective of citizens’ data than those countries, are now looking at a similar approach, using apps and cellphone data.

“I think that everything is gravitating towards proximity tracking,” said Chris Boos, a member of Pan-European Privacy-Preserving Proximity Tracing, a project that is working to create a shared system that could take uploads from apps in different countries. “If somebody gets sick, we know who could be infected, and instead of quarantining millions, we’re quarantining 10.” ...

Some European countries are going further, creating programs to help track individuals—with their permission—who have been exposed and must be quarantined. The Czech Republic and Iceland have introduced such programs and larger countries including the U.K., Germany and Spain are studying similar efforts. Hundreds of new location-tracking apps are being developed and pitched to those governments, Mr. Boos said.

U.S. authorities are able to glean data on broad population movements from the mobile-marketing industry, which has geographic data points on hundreds of millions of U.S. mobile devices, mainly taken from apps that users have installed on their phones.

Europe’s leap to collecting personal data marks a shift for the continent, where companies face more legal restrictions on what data they may collect. Authorities say they have found workarounds that don’t violate the European Union’s General Data Protection Regulation, or GDPR, which restricts how personal information can be shared. ...

Google, Apple, Facebook, etc. are reluctant to draw attention to their already formidable geolocation capabilites. But this crisis may focus public awareness on their ability to track almost all Americans throughout the day.

WSJ: Google will help public health officials use its vast storage of data to track people’s movements amid the coronavirus pandemic, in what the company called an effort to assist in “unprecedented times.”

The initiative, announced by the company late Thursday, uses a portion of the information that the search giant has collected on users, including through Google Maps, to create reports on the degree to which locales are abiding by social-distancing measures. The “mobility reports” will be posted publicly and show, for instance, whether particular localities, states or countries are seeing more or less people flow into shops, grocery stores, pharmacies and parks. ... 

This is just a hint at what Google is capable of. Check out Google Timeline! Of course, users have to opt in to create their Google Timeline. But it should be immediately obvious that Google already HAS the information necessary to populate a detailed geolocation history of every individual...




Added from the comments:

There are really two separate issues here:

1. What is the basic epidemiology of CV19? i.e., R0, CFR, age distribution of vulnerability, comorbidities, mechanism of spread, utility of masks, etc.

2. What is the cost benefit analysis for various strategies (e.g., lockdown vs permissive sweep with cocooning)

While we have not reached full convergence on #1 I think reasonable people agree that the "mainstream" consensus has a decent chance of being correct: e.g., CFR ~ 1% or so, possibility of wide sweep in any population, overload of ICUs means much higher CFR, warmer weather might not save the day, etc. Once this scenario for #1 has, say, >50% chance of being right you are forced to at least take it seriously and then you are on to #2. (It is not required to believe that the scenario above is true at 95% or 99% confidence level...)

#2 is a question of trade-offs and two reasonable people can easily disagree until the end of time... I've already posted very simple CBA that show the answer can go either way depending on how you "price" QALYs, what you think long term effects on economy are from lockdown -- i.e., how fragile you think financial, supply chain, psychological systems are in various places; is it a ~$trillion cost, or could it go nonlinear?

Re: Physicists (and addressing gmachine comment below which has a lot of truth in it), we have no trouble understanding modeling done by other people (whether in finance, climate, or epidemiology), and we are also trained to deal with very uncertain data / statistical situations. We can "take apart" the model in our head to see where the dependencies are and how the uncertainties propagate through the model. I am amazed often to meet people who built a very complex model (e.g., thousands of lines of code, lots of input parameters), but they lack the chops to develop good intuition for how their model works, to make qualitative estimates for uncertainty quantification, etc. I have seen this in economics, finance, biology, and climate contexts many times. "There are levels to this thing..." Understanding the model can be more g-loaded than building it!

Finally, we are trained to think from first principles -- which assumptions are crucial to reach the conclusions, which are not? What are the key uncertainties in the analysis? Do we really need very specific assumptions about, e.g., social interaction rates as in the Imperial models? Or can I do a quick Fermi estimate which gets me a more robust answer at the cost of a factor of 2 uncertainty that does not really affect the main conclusion -- e.g., will ICU overload happen?

Enrico Fermi at the Trinity test: "I tried to estimate its strength by dropping from about six feet small pieces of paper before, during, and after the passage of the blast wave. Since, at the time, there was no wind I could observe very distinctly and actually measure the displacement of the pieces of paper that were in the process of falling while the blast was passing. The shift was about 2 1/2 meters, which, at the time, I estimated to correspond to the blast that would be produced by ten thousand tons of T.N.T." The actual yield was about 20 kt. Sometimes a smart guy can get to within a factor of two, and with much greater clarity, than a huge team of modelers...

Klaus Lackner on Carbon Capture, Climate Change, and Physics

Steve and Corey talk to Klaus Lackner, director of the Center for Negative Carbon Emissions (CNCE) at Arizona State University and the first person to suggest removing CO2 from air to address climate change. Steve asks whether Klaus’ research was motivated by a tail risk of catastrophic outcomes due to CO2 build up. Klaus explains that he sees atmospheric CO2 as a waste management problem. Calculations show that removing human-produced carbon is energetically and economically viable. Klaus describes his invention, a “mechanical tree”, that passively collects CO2 from the air, allowing it to be stored or converted to fuel.

Note: Klaus, in theorist fashion, performs a number of Fermi estimates in real time during the discussion. To fully understand his reasoning, it might be useful to consult the transcript or replay the relevant parts of the interview.

Transcript

Klaus Lackner (Faculty Bio)

Center for Negative Carbon Emissions at ASU


man·i·fold /ˈmanəˌfōld/ many and various.

In mathematics, a manifold is a topological space that locally resembles Euclidean space near each point.

Steve Hsu and Corey Washington have been friends for almost 30 years, and between them hold PhDs in Neuroscience, Philosophy, and Theoretical Physics. Join them for wide ranging and unfiltered conversations with leading writers, scientists, technologists, academics, entrepreneurs, investors, and more.

Steve Hsu is VP for Research and Professor of Theoretical Physics at Michigan State University. He is also a researcher in computational genomics and founder of several Silicon Valley startups, ranging from information security to biotech. Educated at Caltech and Berkeley, he was a Harvard Junior Fellow and held faculty positions at Yale and the University of Oregon before joining MSU.

Corey Washington is Director of Analytics in the Office of Research and Innovation at Michigan State University. He was educated at Amherst College and MIT before receiving a PhD in Philosophy from Stanford and a PhD in a Neuroscience from Columbia. He held faculty positions at the University Washington and the University of Maryland. Prior to MSU, Corey worked as a biotech consultant and is founder of a medical diagnostics startup.

Show Website

COVID-19: the weeks ahead

The figures below (#deaths) are from the Financial Times (data updates frequently). It's useful to compare NY and US numbers to Lombardia and Italy. In the former case we can extrapolate the slope on the log plot for another week or so, because lockdowns here are recent and death generally happens +2-3wks after infection. For Lombardia / Italy we can start to see some bending of the curve from lockdowns that started about 20 days ago.

Based on these simple observations, I think NY will be ~10k in a week, US perhaps double that. Comparing with the Italian curves, US would be lucky not to reach ~100k by end of April -- i.e. 4 or 5 doublings in next 4 wks, or 2.5k x (16-32) = 40-80k.

In COVID-19 Notes (March 8) I gave my best estimates for key parameters characterizing the epidemic:

1. R0 ~ (2-3) or higher in a permissive environment -- no strong efforts at social distancing, quarantine, etc.

2. Fatality rate: roughly 1 percent of cases, heavily concentrated in older individuals and/or those with pre-existing conditions. Note this assumes a well-functioning health system and resources for the 5% or so of cases that need intensive care. See below.

3. In situations like #1 above, doubling time could be as short as a few days. Number of infections in Italy grew by ~1000x over the month of February -- i.e., 2^10 or 2+ doublings per week!

80 percent mild case
15 percent serious (may require hospitalization)
5 percent ICU

I think these numbers will turn out to be roughly correct. For the cognoscenti: German CFR numbers have been steadily increasing, just under 1% now, SK increasing to about 1.5%. Diamond Princess deaths have reached 10, perhaps more to come.

This is from a widely-circulated post by a New Orleans ER doc:

I am an ER MD in New Orleans. Class of 98 [Texas A&M]. Every one of my colleagues have now seen several hundred Covid 19 patients and this is what I think I know.

Clinical course is predictable.
2-11 days after exposure (day 5 on average) flu like symptoms start. Common are fever, headache, dry cough, myalgias(back pain), nausea without vomiting, abdominal discomfort with some diarrhea, loss of smell, anorexia, fatigue.

Day 5 of symptoms- increased SOB, and bilateral viral pneumonia from direct viral damage to lung parenchyma.

Day 10- Cytokine storm leading to acute ARDS and multiorgan failure. You can literally watch it happen in a matter of hours.

[ NOTE: (2-11) + 10 = 2-3 wks after infection ]

81% mild symptoms, 14% severe symptoms requiring hospitalization, 5% critical.

... Our main teaching hospital repurposed space to open 50 new Covid 19 ICU beds this past Sunday so these numbers are with significant decompression. Today those 50 beds are full. They are opening 30 more by Friday. But even with the "lockdown", our AI models are expecting a 200-400% increase in covid 19 patients by 4/4/2020.

... Everyone is scared; patients and employees. But we are the leaders of that emergency room. Be nice to your nurses and staff. Show by example how to tackle this crisis head on. Good luck to us all.

Judging from the video below, I would guess that there are at least ~1 million people in NYC among whom the infection rate has been rapid (~3 day doubling timescale) for a long time, with little to no reduction due to the lockdown. If this is correct, we can expect a huge number of deaths and critical cases from this population alone -- perhaps 10k deaths, 50k critical cases.




Added: This is an interview with one of the leading CV-19 experts in Korea. Not much new for readers of this blog, but a good introduction for the general population. Perhaps most interesting: discussion of masks at @15m (droplet and aerosol transmission just before that), followed by the Korean approach to the epidemic. Alarmingly, @8m he describes known cases of individual re-infection after recovery!

COVID-19, Blockchain, and the Global Startup Scene - Manifold Podcast #39

Steve and Corey talk to Kieren James-Lubin and Victor Wong of the blockchain technology startup, BlockApps. They begin with a discussion of the COVID-19 epidemic (~25m): lockdown, predictions of ICU overload, and helicopter money. Will personal contact tracking become the new normal? Transitioning to blockchain, a technology many view as viable even in times of widespread societal disruption, they give a basic explanation of the underlying cryptographic and consensus algorithms. Kieren and Victor explain how BlockApps was founded, its business model, and history as a startup. They conclude with a comparison of startup ecosystems in China, Silicon Valley, and NYC.

Recorded on March 18, 2020. Now (March 26) I feel we can make much stronger predictions about CV-19 in the US. We will definitely see overloaded health systems (ICUs) across a broad part of the country. It is already starting to happen in NYC. I will be surprised if the US can avoid tens of thousands of fatalities by early April (say, 14 days from today).

1:08 - Lockdown and ICU Overload COVID-19
5:22 - Singapore and Taiwan Response
17:28 - Government Intervention and Helicopter Money
22:13 - End of the Lockdown?

25:58 - How BlockApps got started
28:56 - Private & Public Key Cryptography and Digital Signatures
34:40 - Blockchain
46:05 - Enterprise Blockchains
1:03:37 - Elevator Pitch
1:24:58 - Global Startup Scene

Transcript

Kieren James-Lubin

Victor Wong


man·i·fold /ˈmanəˌfōld/ many and various.

In mathematics, a manifold is a topological space that locally resembles Euclidean space near each point.

Steve Hsu and Corey Washington have been friends for almost 30 years, and between them hold PhDs in Neuroscience, Philosophy, and Theoretical Physics. Join them for wide ranging and unfiltered conversations with leading writers, scientists, technologists, academics, entrepreneurs, investors, and more.

Steve Hsu is VP for Research and Professor of Theoretical Physics at Michigan State University. He is also a researcher in computational genomics and founder of several Silicon Valley startups, ranging from information security to biotech. Educated at Caltech and Berkeley, he was a Harvard Junior Fellow and held faculty positions at Yale and the University of Oregon before joining MSU.

Corey Washington is Director of Analytics in the Office of Research and Innovation at Michigan State University. He was educated at Amherst College and MIT before receiving a PhD in Philosophy from Stanford and a PhD in a Neuroscience from Columbia. He held faculty positions at the University Washington and the University of Maryland. Prior to MSU, Corey worked as a biotech consultant and is founder of a medical diagnostics startup.

COVID-19: NYC and US estimates

NYC analysis: total deaths (red line) are growing at ~10x per week. Because death typically occurs at least +14 days after infection, the behavior of this curve (doubling timescale, or slope of log plot) is mostly baked in for the next two weeks. It depends on the trailing rate of growth of infections, which has not slowed very much, except perhaps in the last week or so.

Only in the last week or less have we seen significant lockdown and isolation in NYC. So I would be very surprised if there were not another factor of 10x growth in total deaths in the next ~10 days. This would mean over a thousand total deaths, and at least roughly five thousand in need of intensive care (i.e., about 5 cases in critical condition for each fatality). This is, I believe, greater than the available ICU capacity in NYC, even including emergency expansion, military hospital ships, etc. (Roughly, 2k base + 1k emergency expansion + 1k hospital ships. Cuomo has talked about this recently.)

With deep sadness, I predict a health system crisis in NYC within a week or so.

The analogous graphs for the US as a whole also have total deaths and total infections growing with a ~3 day doubling timescale. The infection numbers are mainly limited by testing, but there is no reason to expect a significant decrease in slope of the red (deaths) curve. There is probably another 10x growth in total fatalities over the next ~14 days that we can no longer affect -- the infections have already happened.

Figures obtained here.

A week may have been too optimistic...

NYT: ... All of the more than 1,800 intensive care units in the city are expected to be full by Friday, according to a Federal Emergency Management Agency briefing obtained by The New York Times. Patients could stay for weeks, limiting space for newly sickened people.

... At least two city hospitals have filled up their morgues, and city officials anticipated the rest would reach capacity by the end of this week, according to the briefing. The city requested 85 refrigerated trailers from FEMA for mortuary services, along with staff, the briefing said.

... Other doctors said they had tried to resuscitate people while drenched in sweat under their protective gear, face masks fogging up. Some patients have been found dead in their rooms while doctors were busy helping others, they said.

COVID-19: Cocoon the vulnerable, save the economy?

1. Define a high-risk (age 65+, pre-existing condition) and low-risk population. Perhaps 15% in the first category, 85% in the latter.

2. Let the low-risk people go back to work. Start the economy up again.

3. Issue a "shelter in place" order for ALL high-risk people. Low-risk individuals who co-habit with high-risk people (e.g., family members) have a choice: self-isolate as well, or move into government provided accommodation.

4. The isolated people receive regular food deliveries and medical care (virus tests). Existing platforms for food delivery (Lyft, Uber) can be used almost immediately for this purpose.

This seems to be a good solution in that:

a. it minimizes damage to the economy,
b. reduces deaths significantly (if executed efficiently), and
c. matches the sacrifices (inconveniences) to the people who benefit most.

Would it work? When I first heard about herd immunity and cocooning the vulnerable, I was skeptical. But now that whole countries have experienced lockdown, and financial markets have descended into chaos, I suspect that the solution above would be widely acceptable as an alternative.

Previous discussion of this idea:

I've yet to see any detailed calculations for an alternative plan which tries to isolate the most vulnerable while allowing the economy to continue to function. Could it work?

1. Define the vulnerable population: e.g., people over 65 or with pre-existing conditions. Let's say this is 15% of the total population.

2. Try to isolate these people as much as possible. Give them free accommodation in the now empty hotels and college dormitories, on military bases and in makeshift shelters. Guarantee them free food delivery and regular medical checkups in these new locations. Within these locations practice very good internal isolation, as on the Princess cruise ship: isolate each person in their room, require the wearing of (inexpensive; not N95) masks when they emerge for activities, etc. Use the National Guard and military to execute the strategy.

Alternatively, we could leave the vulnerable population in place where they are now and move out any younger people (e.g., family members) they live with into temporary housing, or even confine the younger people in place if they consent to isolation in order to help their elders. The shelter in place order only applies to 65+ instead of the whole population!

It seems that this could be less expensive than the lockdown currently in place.

Uber and Lyft drivers wearing gloves and masks, spraying down their cars periodically, could be used as people movers or for food delivery. Instead of shutting down restaurants, the government could employ them to produce the meals for the sequestered people... The government just has to pay Uber and Lyft through their existing platform. Beats mailing checks to the entire population!

Diamond Princess data: "3,711 passengers and crew members onboard. Passengers were initially to be held in quarantine for 14 days. However, those that had intense exposure to the confirmed case-patient, such as sharing a cabin, were held in quarantine beyond the initial 14-day window [3]. By 20th February, there were 634 confirmed cases onboard (17%), with 328 of these asymptomatic (asymptomatic cases were either self-assessed or tested positive before symptom onset) [3]. Overall 3,063 PCR tests were performed among passengers and crew members." Among the infected were many younger passengers and crew members, but the 7 individuals who died were all 70+. AFAIK 10-20 more individuals were in critical condition and we don't know their ultimate fate, but clearly the worst outcomes are highly concentrated among older people. This is reflected in the data from Italy, PRC, and S. Korea as well.

Note: as of March 24 the number of deaths has reached 10, but I don't know the age distribution. In recent NYC data it looks like ~80% are over 60 and the rest a bit younger.

COVID-19: CBA, CFR, Open Borders

Some observations related to COVID-19.

1. Cost-Benefit Analysis: I don't favor this kind of purely economic analysis when human lives are at stake, but at the crudest level we are talking about US policy choices (i.e., lockdown vs permissive sweep) that could lead to of order a million more or fewer deaths. EPA sometimes values a human life at ~$7M, which suggests that the last 10-20 years of life for an older person could be valued at ~$1M. Under these assumptions the economic value of a lockdown which saves a million lives is in the trillion dollar range. This is similar to the cost of shutting down big chunks of the economy for months. So a CBA could go either way depending on detailed assumptions.

2. Open Borders: Many areas of the world are likely to develop into reservoirs for COVID-19. When the US emerges from lockdown it will be in our interests to carefully test (and potentially quarantine) everyone entering the country. I believe Singapore already has such a regime in place, with a reported 3 hour turn around for test results at the airport!

How can we possibly exit lockdown without the kind of control that Singapore exerts? In the COVID-19 reality it is crazy for the US not to impose strict control over the southern border. This will impact the presidential campaign in interesting ways. I don't think it will be tenable for a candidate to tell the virus-weary citizenry that it is impossible to control entry into the country. Once the infrastructure is built to effectively control our borders, it will be difficult to dismantle.

3. Case Fatality Rate: What is the lower bound on CFR (or more precisely, IFR = Infected Fatality Rate)? I have been working under the assumption that about 1% of infected individuals will die from COVID-19, assuming access to good medical care. (IIRC the value from S. Korea data is just below 1%.) I think there is a factor of 2-3 uncertainty in this number, so it could be as low as 0.5 or even 0.3%. Lower numbers are typically justified by invoking many asymptomatic, undetected cases in the population. However, large numbers of undetected infections also imply very rapid spread of the disease.

Rapid spread (e.g., in the absence of a rigorous lockdown) would lead to an overloaded medical system, with a resulting IFR which could be several times higher than the value that applies under good conditions -- so 0.3% could become 1%. In a scenario in which the virus sweeps through most of the population in a relatively short time (e.g., 3-6 months), an overall death rate of 1% times, e.g., 50-70% of the population is still a reasonable estimate. For the USA this would be at least 1.5 million people.

Related posts on COVID-19.

Hospitals in NYC already under stress: WSJ , NYT


Note Added: I've yet to see any detailed calculations for an alternative plan which tries to isolate the most vulnerable while allowing the economy to continue to function. Could it work?

1. Define the vulnerable population: e.g., people over 65 or with pre-existing conditions. Let's say this is 15% of the total population.

2. Try to isolate these people as much as possible. Give them free accommodation in the now empty hotels and college dormitories, on military bases and in makeshift shelters. Guarantee them free food delivery and regular medical checkups in these new locations. Within these locations practice very good internal isolation, as on the Princess cruise ship: isolate each person in their room, require the wearing of (inexpensive; not N95) masks when they emerge for activities, etc. Use the National Guard and military to execute the strategy.

Alternatively, we could leave the vulnerable population in place where they are now and move out any younger people (e.g., family members) they live with into temporary housing, or even confine the younger people in place if they consent to isolation in order to help their elders. The shelter in place order only applies to 65+ instead of the whole population!

It seems that this could be less expensive than the lockdown currently in place.

Uber and Lyft drivers wearing gloves and masks, spraying down their cars periodically, could be used as people movers or for food delivery. Instead of shutting down restaurants, the government could employ them to produce the meals for the sequestered people... The government just has to pay Uber and Lyft through their existing platform. Beats mailing checks to the entire population!

COVID-19: Smart Technologies and Exit from Lockdown (Singapore)

How will we exit lockdown? Once the rate of spread of COVID-19 has been reduced sufficiently, we should be able to switch to a more precise, targeted methodology. (Of course, widespread testing is a prerequisite baseline capability...) Below are some methods in use in Singapore.



This is the most benign app for contact tracing I can imagine. But it requires a large fraction of people to be running it on their phones. Geo-location capabilities are strong enough that most governments could do this without the opt-in app. They should prepare a legal consent form for future use that permits aggressive contact tracing using all data available (e.g., from Google or Apple or cell carriers). Most people who are diagnosed with CV-19 would probably consent to the use of their geo-location data to help others they may have infected.

Here's a simple system for enforcing home quarantine. If the UK and US governments can't deploy something like this to help us exit from lockdown in a month or two, then our civilization is toast!

The simplest technology of all is to get everyone to wear a mask in public. Not a fancy N95 mask - even a home made paper or cloth mask would do to stop droplet spray. This is likely as effective as social distancing in stopping virus spread. I suspect widespread use of masks may have helped countries like Japan and Taiwan significantly.

People who are bemoaning the economic costs of this lockdown should be very focused on technologies and behaviors that will allow us to exit as soon as possible, while keeping the epidemic under control.

Claude Steele on the Challenges of Multi-Cultural Societies - Manifold Podcast #38

Corey and Steve talk to Claude Steele of Stanford about his article Why Campuses are So Tense. The essay explores stereotype threats across racial lines. Colorblindness is a standard of fairness, but what are the costs of ignoring our differences? Claude describes his research on minority under-performance and why single sex colleges may contribute to women’s success. Corey describes why he believes his daughter's experience is a counterexample to the findings of the experiments that led the Supreme Court to outlaw segregation. The three discuss parenting in a diverse world and how ethnic integration differs between Europe and the US.

Transcript

Claude Steele

Why Are Campuses So Tense?

In Struggle: SNCC and the Black Awakening of the 1960s


man·i·fold /ˈmanəˌfōld/ many and various.

In mathematics, a manifold is a topological space that locally resembles Euclidean space near each point.

Steve Hsu and Corey Washington have been friends for almost 30 years, and between them hold PhDs in Neuroscience, Philosophy, and Theoretical Physics. Join them for wide ranging and unfiltered conversations with leading writers, scientists, technologists, academics, entrepreneurs, investors, and more.

Steve Hsu is VP for Research and Professor of Theoretical Physics at Michigan State University. He is also a researcher in computational genomics and founder of several Silicon Valley startups, ranging from information security to biotech. Educated at Caltech and Berkeley, he was a Harvard Junior Fellow and held faculty positions at Yale and the University of Oregon before joining MSU.

Corey Washington is Director of Analytics in the Office of Research and Innovation at Michigan State University. He was educated at Amherst College and MIT before receiving a PhD in Philosophy from Stanford and a PhD in a Neuroscience from Columbia. He held faculty positions at the University Washington and the University of Maryland. Prior to MSU, Corey worked as a biotech consultant and is founder of a medical diagnostics startup.

COVID-19: Some US estimates -- the knife's edge

My lower bound on current number of cases in the US (March 18): tens of thousands. Positive test results are just the tip of the iceberg.

The unconstrained doubling time is roughly 3 days. Some recent estimates of this timescale by country (note these are trailing estimates): Japan is ~10 days, S. Korea ~16, PRC ~36 days currently. Most European countries are at 3-4 days, although these estimates are unreliable as the increase is mainly due to wider testing and does not capture growth rate very well.

In Michigan we've closed K-12 schools and universities, as well as bars, restaurants, gyms, etc. I would guess that the doubling time has been increased to ~7 or 10 days by these measures.

Medical system overload will happen for sure at ~1 million total cases in the US, based on 5% of cases requiring intensive care, and total ICU capacity (~50k) in the US. Local regions may go into crisis much earlier, in particular dense urban regions that are more cosmopolitan than the US as a whole. The threshold is roughly 0.3 percent of population infected -- only a few per thousand!

Overload may happen much earlier -- hospital systems run at over 90% occupancy. So most spaces are already filled by sick people. The numbers above might be optimistic by an order of magnitude -- perhaps 100k total cases in US, or 0.03 percent of local population infected will lead to at least some health system crises.

Another factor contributing to early overload is exhaustion of protective equipment for medical personnel (already starting to happen in some places like WA) and eventual sickness of those personnel.

Best case: Widespread US crisis after another (3-5) doublings:  2^3 to 2^5 more cases. Assuming the whole country adopts isolation measures as in Michigan: roughly 30 days until widespread crises. I suspect we will see localized crises much sooner.

How might things turn out better?

Warm weather plus isolation measures push doubling time to 14 or even 20 days. Even in the best case this won't apply throughout the whole country, so still expect problems in some regions.

Earlier posts: COVID-19 Notes , COVID-19 Update: Developing Countries and Flattening the US Curve.


The figures below are from this NYTimes article describing a Harvard analysis similar to the one above. Note in the first figure the solid yellow line is crossed in all scenarios very early on -- i.e., +1 or 2 months. Overloaded ICUs are likely in the near future.

COVID-19 Update: Developing Countries and Flattening the US Curve

SITUATION IN THE PHILIPPINES:

One of the key unknowns about this epidemic is the status of developing countries, for example those in warm equatorial regions such as in Africa and India. If we are fortunate, we will find that warm weather and humidity inhibit the spread of COVID-19. But this is not yet conclusively established.

In the video below Dr. John Campbell describes the situation in the Philippines (population 100M): hospitals are overwhelmed in Manila (population 20M) and drastic measures have been taken by the government. Given the lack of medical infrastructure, the mortality rate there could be well above 1% for cases. This would mean millions of deaths if the virus sweeps the population.

Even worse, we might expect the epidemic to play out similarly among billions of other people around the world. Perhaps climactic protection does not extend to Manila for specific reasons, such as the widespread use of air conditioning, the large malls and food courts, etc. (What is happening in Bangkok and Kuala Lumpur?) Or perhaps Manila is a harbinger of what is to come in many other countries.

The Philippines case study might give important clues as to the seasonality of COVID-19. How much relief will summer provide?





FLATTENING THE CURVE IN THE US:

The calculations below are from an email I sent yesterday, describing what is required to "flatten the curve" in the US. The dotted line in the figure above can be interpreted as roughly 0.3 percent infected in any local population region (e.g., NYC or Seattle). For the US as a whole, this is ~1 million total infections in the country at any given time.

By my estimates the US health system will be overwhelmed if we have to deal with ~1 million cases (which is only 0.3% of the population) at any instant of time. This comes from counting free ICU space (~50k beds in US) and the observed fact that ~5% of cases require intensive care. See more here:

https://infoproc.blogspot.com/2020/03/covid-19-notes.html

If the number of infected at any given time is higher than ~1M the mortality rate will be higher than the ~1% observed under good medical conditions. Even under good conditions, where the 1% applies: 50% of US population infected (i.e., some kind of sweep) means 1.5 million deaths! This could easily be 2 or 3 times higher if our health system is overwhelmed, as in northern Italy and Iran.

Now, assuming an ICU treatment period of 7 days (optimistic?), the amount of time required to deal with 1 million infected at a time on the way to an integrated 150 million (i.e., 50% infected) is 7 x 150 days = 150 weeks... about 3 years! Obviously during this period we would build more intensive care infrastructure, etc. but the point remains that flattening the curve so that we avoid the situation where people who need intensive care are turned away, causing death rate among cases to shoot well above 1%, will be challenging!

I think aggressive measures are absolutely necessary to avoid a catastrophic outcome. Whether, e.g., K-12 closures have a material impact on that is not completely clear to me, but I think we should probably err on the side of caution.

We have to reduce R0 drastically before hitting ~1M cases or we are in for mass panic. In February Italian cases grew by 1000x. That means we have only weeks to get things under control in US before we overwhelm our health system.

The main unknown: How effective are the measures now widely taken in the US (schools and companies moving to remote operation, expanded testing and quarantine) going to be in reducing the doubling rate from 2+ times per week (i.e., doubling time of few days, as observed in Italy in February) to something less rapid?

We are going to find out in coming weeks.

A.J. Robison on the Neural Basis of Sex Differences in Depression - Manifold #37

Corey and Steve talk with MSU Neuroscientist A.J. Robison about why females may be more likely to suffer from depression than males. A.J. reviews past findings that low testosterone and having a smaller hippocampus may predict depression risk. He explains how a serendipitous observation opened up his current line of research and describes tools he uses to study neural circuits. Steve asks about the politics of studying sex differences and tells of a start up using CRISPR to attack heart disease. The three end with a discussion of the psychological effects of ketamine, testosterone and deep brain stimulation.

01:18 - Link between antidepressants, neurogenesis and reducing risk of depression

13:54 - Nature of Mouse models

23:19 - Depressive symptoms in mouse

32:36 - Liz Williams' serendipitous finding and the issue of biological sex

45:47 - AJ’s research plans for circuit specific gene editing in the mouse brain and a start up’s plan to use it to tackle human cardiovascular disease

59:07 - Psychological and Neurological Effects of Ketamine. Testosterone and Deep Brain Stimulation

Transcript

Robison Lab at MSU

Androgen-dependent excitability of mouse ventral hippocampal afferents to nucleus accumbens underlies sex-specific susceptibility to stress

Emerging role of viral vectors for circuit-specific gene interrogation and manipulation in rodent brain


man·i·fold /ˈmanəˌfōld/ many and various.

In mathematics, a manifold is a topological space that locally resembles Euclidean space near each point.

Steve Hsu and Corey Washington have been friends for almost 30 years, and between them hold PhDs in Neuroscience, Philosophy, and Theoretical Physics. Join them for wide ranging and unfiltered conversations with leading writers, scientists, technologists, academics, entrepreneurs, investors, and more.

Steve Hsu is VP for Research and Professor of Theoretical Physics at Michigan State University. He is also a researcher in computational genomics and founder of several Silicon Valley startups, ranging from information security to biotech. Educated at Caltech and Berkeley, he was a Harvard Junior Fellow and held faculty positions at Yale and the University of Oregon before joining MSU.

Corey Washington is Director of Analytics in the Office of Research and Innovation at Michigan State University. He was educated at Amherst College and MIT before receiving a PhD in Philosophy from Stanford and a PhD in a Neuroscience from Columbia. He held faculty positions at the University Washington and the University of Maryland. Prior to MSU, Corey worked as a biotech consultant and is founder of a medical diagnostics startup.

COVID-19 Notes

First some basic assumptions, for which I think the evidence is strong (reference):

1. R0 ~ (2-3) or higher in a permissive environment -- no strong efforts at social distancing, quarantine, etc.

2. Fatality rate: roughly 1 percent of cases, heavily concentrated in older individuals and/or those with pre-existing conditions. Note this assumes a well-functioning health system and resources for the 5% or so of cases that need intensive care. See below.

3. In situations like #1 above, doubling time could be as short as a few days. Number of infections in Italy grew by ~1000x over the month of February -- i.e., 2^10 or 2+ doublings per week!

USA has perhaps ~1M total hospital beds, over half already occupied, and perhaps 50k ICU spaces. For those infected, the distribution of severity is roughly (again, concentration in vulnerable sub-populations):

80 percent mild case
15 percent serious (may require hospitalization)
5 percent ICU

So roughly 1M infected at a given time would overwhelm US health capabilities. We probably have at least ~1000 infected in the country at the moment, so in the absence of serious measures like social distancing (cancellation of sporting events, large meetings, moving to K12 and college distance learning, etc.), we would reach the health system breaking point in about a month. Many other countries, in Europe and elsewhere, are facing a similar situation.

Whether we impose draconian social measures (which would have a strongly negative effect on cafes, restaurants, hotels, airlines, theaters, etc.) or let COVID-19 infect millions of people, we are in for at least a one quarter downturn (recession?) with the possibility of more significant nonlinear events (complete market collapse, systemic failures). Traders already understand this, which is why equities are in huge decline despite a 50 bp Fed rate cut last week. I went largely to cash already...

We have technology that could help us fight the epidemic. The article below, in the Journal of the American Medical Association, describes how Taiwan successfully handled the epidemic -- less than 50 cases! -- despite close proximity and extensive travel to China. (Note, Taiwan in Jan-Feb is a bit warmer than Milan, but I don't think climate is the entire reason for their good performance...) Google and Apple have these technical (geolocation, tracking) capabilities, but they don't like to emphasize it to the public.

Response to COVID-19 in Taiwan: Big Data Analytics, New Technology, and Proactive Testing

JAMA. Published online March 3, 2020. doi:10.1001/jama.2020.3151

Taiwan is 81 miles off the coast of mainland China and was expected to have the second highest number of cases of coronavirus disease 2019 (COVID-19) due to its proximity to and number of flights between China.1 The country has 23 million citizens of which 850 000 reside in and 404 000 work in China.2,3 In 2019, 2.71 million visitors from the mainland traveled to Taiwan.4 As such, Taiwan has been on constant alert and ready to act on epidemics arising from China ever since the severe acute respiratory syndrome (SARS) epidemic in 2003. Given the continual spread of COVID-19 around the world, understanding the action items that were implemented quickly in Taiwan and assessing the effectiveness of these actions in preventing a large-scale epidemic may be instructive for other countries.

COVID-19 occurred just before the Lunar New Year during which time millions of Chinese and Taiwanese were expected to travel for the holidays. Taiwan quickly mobilized and instituted specific approaches for case identification, containment, and resource allocation to protect the public health. Taiwan leveraged its national health insurance database and integrated it with its immigration and customs database to begin the creation of big data for analytics; it generated real-time alerts during a clinical visit based on travel history and clinical symptoms to aid case identification. It also used new technology, including QR code scanning and online reporting of travel history and health symptoms to classify travelers’ infectious risks based on flight origin and travel history in the past 14 days. Persons with low risk (no travel to level 3 alert areas) were sent a health declaration border pass via SMS (short message service) messaging to their phones for faster immigration clearance; those with higher risk (recent travel to level 3 alert areas) were quarantined at home and tracked through their mobile phone to ensure that they remained at home during the incubation period.

Google + Apple + cell carriers have the data to detect near collisions (proximity) between COVID-19 spreaders (once diagnosed) and other individuals. This can be done anonymously: i.e., public health services get a warning that a spreader visited nursing home X on timestamp T without naming the spreader. Another alternative (see WSJ video below) is to have an OPT-IN app that checks location history and warns if you were in close proximity to a spreader. The PRC government version of this app has been used by 200M+ Chinese already -- note it's OPT-IN. The companies above have the data to do this but don't like it to be known that they can.




Note Added: Latest results from the beginnings of broader testing in Seattle suggest that the virus is widespread in the community already. The number of cases detected is primarily limited by number of tests:

Nature: “We are past the point of containment,” says Helen Chu, an infectious-disease specialist at the University of Washington School of Medicine (UW Medicine) in Seattle. “So now we need to keep the people who are vulnerable from getting sick.”

I checked the weather records for Seattle in February 2020: daily highs in the mid-40s to mid-50s. There is very little chance that wintry areas of the US will be warmer than this over the next 30 days, even with an early spring. So I don't see that US spread in those regions can be mitigated by anything less than social distancing and other strong measures. Weather will probably not save us.

Front line tweets from Italian doctors describe medical resources pushed to the breaking point. I hope we do not experience this in the US, but I don't see how we will avoid it: [1] [2]

Some interesting information about transmission (surfaces, air in confined spaces) at the beginning; Italian overload at 17m; S. Korean data at 19m.

Within-Family Validation of Polygenic Risk Scores and Complex Trait Prediction (bioRxiv)

We examined the performance of disease risk and complex trait predictors using about 40k siblings in late adulthood. As the figures show, predictive power is only modestly reduced in within-family designs, suggesting that we are dealing with real causal genetic effects.

Within-Family Validation of Polygenic Risk Scores and Complex Trait Prediction

Louis Lello, Timothy Raben, Stephen D. H. Hsu

doi: https://doi.org/10.1101/2020.03.04.976654

We test a variety of polygenic predictors using tens of thousands of genetic siblings for whom we have SNP genotypes, health status, and phenotype information in late adulthood. Siblings have typically experienced similar environments during childhood, and exhibit negligible population stratification relative to each other. Therefore, the ability to predict differences in disease risk or complex trait values between siblings is a strong test of genomic prediction in humans. We compare validation results obtained using non-sibling subjects to those obtained among siblings and find that typically most of the predictive power persists in within-family designs. In the case of disease risk we test the extent to which higher polygenic risk score (PRS) identifies the affected sibling, and also compute Relative Risk Reduction as a function of risk score threshold. For quantitative traits we examine between-sibling differences in trait values as a function of predicted differences, and compare to performance in non-sibling pairs. Example results: Given 1 sibling with normal-range PRS score (less than 84th percentile) and 1 sibling with high PRS score (top few percentiles), the predictors identify the affected sibling about 70-90 percent of the time across a variety of disease conditions, including Breast Cancer, Heart Attack, Diabetes, etc. For height, the predictor correctly identifies the taller sibling roughly 80 percent of the time when the (male) height difference is 2 inches or more.

From the paper:

If a girl grows up to be taller than her sister, with whom she spent the first 18 years of her life, it seems likely at least some of the height difference is due to genetic differences. How much of phenotype difference can we predict from DNA alone? If one of the sisters develops breast cancer later in life, how much of the risk was due to genetic variants that she does not share with her asymptomatic sister? These are fundamental questions in human biology, which we address (at least to some extent) in this paper.

... We emphasize that predictors trained on even larger datasets will likely have significantly stronger performance than the ones analyzed here [13, 14]. As we elaborated in earlier work, where many of these predictors were first investigated, their main practical utility at the moment is in the identification of outliers who may be at exceptionally high (or low) risk for a specific disease condition. The results here confirm that high risk score outliers are indeed at elevated risk, even compared to their (normal range score) siblings.

The sibling results presented in this paper, together with the many out of sample validations of polygenic scores that continue to appear in the literature, suggest that genomic prediction in humans is a robust and important advance that will lead to improvements in translational medicine as well as deep insights into human genetics.

Kaja Perina on the Dark Triad: Narcissism, Machiavellianism, and Psychopathy - Manifold Podcast #36

Kaja Perina is the Editor in Chief of Psychology Today. Kaja, Steve, and Corey discuss so-called Dark Triad personality traits: Narcissism, Machiavellianism, and Psychopathy. Do these traits manifest more often in super successful people? What is the difference between Sociopathy and Psychopathy? Are CEOs often "warm sociopaths"? Can too much empathy be a liability? Corey laments Sociopathy in academic Philosophy. Kaja explains the operation of Psychology Today. Steve reveals his Hypomania diagnoses.

2:33 - Psychopathology and the Dark Triad
11:34 - Do these traits manifest more often in super successful people?
17:52 - Can too much empathy be a liability?
35:16 - Corey laments Sociopathy in academic Philosophy
50:32 - Kaja explains the operation of Psychology Today
1:01:06 - Steve reveals his Hypomania diagnoses

Transcript

Kaja Perina (Psychology Today)

Related: Nice Guys Finish Last (2012 post), more Hypomania


man·i·fold /ˈmanəˌfōld/ many and various.

In mathematics, a manifold is a topological space that locally resembles Euclidean space near each point.

Steve Hsu and Corey Washington have been friends for almost 30 years, and between them hold PhDs in Neuroscience, Philosophy, and Theoretical Physics. Join them for wide ranging and unfiltered conversations with leading writers, scientists, technologists, academics, entrepreneurs, investors, and more.

Steve Hsu is VP for Research and Professor of Theoretical Physics at Michigan State University. He is also a researcher in computational genomics and founder of several Silicon Valley startups, ranging from information security to biotech. Educated at Caltech and Berkeley, he was a Harvard Junior Fellow and held faculty positions at Yale and the University of Oregon before joining MSU.

Corey Washington is Director of Analytics in the Office of Research and Innovation at Michigan State University. He was educated at Amherst College and MIT before receiving a PhD in Philosophy from Stanford and a PhD in a Neuroscience from Columbia. He held faculty positions at the University Washington and the University of Maryland. Prior to MSU, Corey worked as a biotech consultant and is founder of a medical diagnostics startup.

Pleiotropy: Myths and Reality

The conventional view on pleiotropy is captured by this excerpt from H. Judson's The Eighth Day of Creation (PDF), p.609:

Genes act in concert. In the innumerable interactions of the developmental process, any single gene may affect several characters. The technical term is pleiotropy. And any character will be influenced by many genes. The term is polygeny. ...

[Thomas Hunt] Morgan's precepts establish a crucial and absolute distinction. While only a single allele, or very few, are involved in any given one of a high proportion of genetically related diseases, general heritable qualities of body or mind, our longevity, our intelligence, our particular talents, are the product of many genes, in exquisite balance among themselves and with the environment. This is why positive eugenics, breeding people for the enhancement of general qualities, cannot work, and why the extreme of positive eugenics, direct inter­vention in the germ line to improve such qualities, is a forbidden experiment: almost certainly you would upset the exquisite balance and engender a new human who would be seriously defective. All genes work in concert.

You will note that this description of pleiotropy lacks quantitative precision. In the time of Morgan: genes were entirely theoretical constructs, the importance and nature of DNA still a mystery, and the idea that there might be billions of distinct genetic variants totally beyond comprehension (perhaps to everyone but Fisher). The concept of pleiotropy was formulated before the notion of high dimensional spaces of variation became familiar.

Leaving aside ethical issues related to genetic enhancement, one can ask the practical scientific question: Is it possible?

In our recent paper (see post Live Long and Prosper: Genetic Architecture of Complex Traits and Disease Risk Predictors), we looked at the extent to which SNPs used in polygenic predictors of risk are correlated across pairs of disease conditions. We found relatively low correlations, as depicted in the table below from the paper.

In the conclusions we wrote:

III. The DNA regions used in disease risk predictors so far constructed seem to be largely disjoint (with a few interesting exceptions), suggesting that individual genetic disease risks are largely uncorrelated.

Observation III has interesting implications for pleiotropy [63–65]. We found that genetic risks are largely uncorrelated for different conditions. This suggests that there can exist individuals with, e.g., low risk simultaneously in each of multiple conditions, for any essentially any combination of conditions. There is no trade-off required between different disease risks ... One could speculate that a lucky individual with exceptionally low risk across multiple conditions might have an unusually long life expectancy.

We can formulate this concretely (operationally?) as follows:

1. Regions of DNA correlated to different disease risks are largely disjoint.

2. It is plausible that causal genetic variants lie in these regions. For example, the predictor SNPs themselves could be causal, or they could tag (be highly correlated in state with) nearby causal variants.

3. Hypothetically, one could edit these causal variants independently, making the beneficiary simultaneously low risk for many conditions. The number of standard deviations of effect size in the polygenic score for each disease that can be modified independently (i.e., without affecting other disease risks or traits) is large and can be directly estimated from our results.

As the figure below (source) makes clear, a few SD change (e.g., ~5 SD, from 99th percentile to 1st percentile) in polygenic score for a given disease risk can lead to a 10x or possibly 100x decrease in absolute probability in having the condition. Our results suggest that the amount of variance available for engineering is much greater than this.

Farewell Freeman Dyson

He was 96 when he passed last Friday, one of the last giants who participated in the creation of Quantum Electrodynamics and modern quantum field theory.

He has appeared many times on this blog. Below are a few links.

The intestinal fortitude of Freeman Dyson: an account of his visit to the University of Oregon.

The evening began ominously. Dyson had a stomach bug -- he declined to eat anything at dinner, and made several emergency trips to the bathroom. After dinner he fell asleep on a couch in the physics building. Facing a packed auditorium, with people sitting in the aisles and filling an adjoining overflow room with video monitor, the other organizers and I decided that we'd offer Freeman the chance to call the whole thing off when we woke him up. Luckily for everyone, he felt much better after the nap, and was obviously energized by the large and enthusiastic crowd. After we finished the Q&A, he turned to me and said "Well, your questions cured my bug!"

Profile in The Atlantic (2010)

The prodigy in question, Freeman Dyson, now middle-aged, stared ahead, his incessant concentration on the road unbroken. ... I asked him whether as a boy he had speculated much about his gift. Had he asked himself why he had this special power? Why he was so bright?

Dyson is almost infallibly a modest and self-effacing man, but tonight his eyes were blank with fatigue, and his answer was uncharacteristic.

“That’s not how the question phrases itself,” he said. “The question is: why is everyone else so stupid?”

I want to emphasize that Dyson was a lovely and gentle person. The answer above is indeed uncharacterstic. But it is true...

Profile in NYTimes Magazine (2009). Below is a sample of work produced by Dyson between the ages of 5 and 9.

From Disturbing the Universe, one of my favorite scientific memoirs. It describes dramatic events of Dyson's early life: childhood in England, the war, QED, Feynman and Oppenheimer.

... In that spring of 1948 there was another memorable event. Hans [Bethe] received a small package from Japan containing the first two issues of a new physics journal. Progress of Theoretical Physics, published in Kyoto. The two issues were printed in English on brownish paper of poor quality. They contained a total of six short articles. The first article in issue No. 2 was called "On a Relativistically Invariant Formulation of the Quantum Theory of Wave Fields," by S. Tomonaga of Tokyo University. Underneath it was a footnote saying, "Translated from the paper . . . (1943) appeared originally in Japanese." Hans gave me the article to read. It contained, set out simply and lucidly without any mathematical elaboration, the central idea of Julian Schwinger's theory. The implications of this were astonishing. Somehow or other, amid the ruin and turmoil of the war, totally isolated from the rest of the world, Tomonaga had maintained in Japan a school of research in theoretical physics that was in some respects ahead of anything existing anywhere else at that time. He had pushed on alone and laid the foundations of the new quantum electrodynamics, five years before Schwinger and without any help from the Columbia experiments. He had not, in 1943, completed the theory and developed it as a practical tool. To Schwinger rightly belongs the credit for making the theory into a coherent mathematical structure. But Tomonaga had taken the first essential Step. There he was, in the spring of 1948, sitting amid the ashes and rubble of Tokyo and sending us that pathetic little package. It came to us as a voice out of the deep.

A few weeks later, Oppy received a personal letter from Tomonaga describing the more recent work of the Japanese physicists. They had been moving ahead fast in the same direction as Schwinger. Regular communications were soon established. Oppy invited Tomonaga to visit Princeton, and a succession of Tomonaga's students later came to work with us at Princeton and at Cornell. When I met Tomonaga for the first time, a letter to my parents recorded my immediate impression of him "He is more able than either Schwinger or Feynman to talk about ideas other than his own. And he has enough of his own too. He is an exceptionally unselfish person." On his table among the physics journals was a copy of the New Testament.

Rest in Peace, Freeman Dyson.

Adam Dynes on Noisy Retrospection: The Effect of Party Control on Policy Outcomes - Manifold #35

Steve and Corey talk to Adam Dynes of Brigham Young University about whether voting has an effect on policy outcomes. Adam’s work finds that control of state legislatures or governorships does not have an observable effect on macroscopic variables such as crime rates, the economy, etc. Possible explanations: parties push essentially the same policies, politicians don't keep promises, monied interest control everything. Are voting decisions just noisy mood affiliation? Perhaps time is better spent obsessing about sports teams, which at least generates pleasure.

1:22 - What is retrospective voting?
5:43 - Research findings on retrospective voting
14:02 - Uniparty/Monied interests?
17:23 - Martin Gilens' research
23:10 - Are people just voting based on noise or mood affiliation?
27:13 - Bryan Caplan - Myth of the Rational Voter
34:35 - Is time better spent obsessing about sports teams, which at least generates pleasure?
39:42 - After the fall of Athens, was democracy commonly referred to as irrational mob rule?
48:22 - Does this research translate to the national level?
52:19 - Super Nerdy Stuff: Statistical Analysis, Reproducibility & Null Results
56:40 - Reactions to the results

Transcript

Adam Dynes (Personal Website)

Adam Dynes (Faculty Profile)

Noisy Retrospection: The Effect of Party Control on Policy Outcomes

Related: 2016 blog post on Martin Gilens' work: American and Chinese Oligarchies.


man·i·fold /ˈmanəˌfōld/ many and various.

In mathematics, a manifold is a topological space that locally resembles Euclidean space near each point.

Steve Hsu and Corey Washington have been friends for almost 30 years, and between them hold PhDs in Neuroscience, Philosophy, and Theoretical Physics. Join them for wide ranging and unfiltered conversations with leading writers, scientists, technologists, academics, entrepreneurs, investors, and more.

Steve Hsu is VP for Research and Professor of Theoretical Physics at Michigan State University. He is also a researcher in computational genomics and founder of several Silicon Valley startups, ranging from information security to biotech. Educated at Caltech and Berkeley, he was a Harvard Junior Fellow and held faculty positions at Yale and the University of Oregon before joining MSU.

Corey Washington is Director of Analytics in the Office of Research and Innovation at Michigan State University. He was educated at Amherst College and MIT before receiving a PhD in Philosophy from Stanford and a PhD in a Neuroscience from Columbia. He held faculty positions at the University Washington and the University of Maryland. Prior to MSU, Corey worked as a biotech consultant and is founder of a medical diagnostics startup.

Cold Spring Harbor Laboratory: Seminar and Photos

Last week I visited Cold Spring Harbor Laboratory to give a seminar.

The new material is in slides 13-17. See also Live Long and Prosper: Genetic Architecture of Complex Traits and Disease Risk Predictors. I believe the sibling validation results are extremely important: typically most of the predictive power persists in within-family validation tests. We have not released this paper but will soon -- the slides are a preview. To be honest I fully anticipated these results: the large number of out of sample predictor validations using unrelated individuals strongly suggests that real genetic effects are at work. However, many people are irrationally biased against -- have strong priors against -- genetic causation of complex traits (even disease risks). These family designs provide important "gold standard" evidence, which, one can hope, will enlighten even the most stubborn. The sad alternative is progress one funeral at a time...

Otherwise the talk is similar to the one I gave at the Berkeley/UCSF Innovative Genomics Institute last summer. Video of IGI talk.

Title: Genomic Prediction of Complex Traits and Disease Risks via AI/ML and Large Genomic Datasets

Abstract: The talk is divided into two parts. The first gives an overview of the rapidly advancing area of genomic prediction of disease risks using polygenic scores. We can now identify risk outliers (e.g., with 5 or 10 times normal risk) for about 20 common disease conditions, ranging from diabetes to heart diseases to breast cancer, using inexpensive SNP genotypes (i.e., as offered by 23andMe). We can also predict some complex quantitative traits (e.g., adult height with accuracy of few cm, using ~20k SNPs). I discuss application of these results in precision medicine as well as embryo selection in IVF, and give some details about genetic architectures. The second part covers the AI/ML used to build these predictors, with an emphasis on "sparse learning" and phase transitions in high dimensional statistics.

Some photos. The ones on the wall of the seminar room capture a golden era in molecular biology and the study of DNA. Leo Szilard on the right in the one below. Also, Jacques Monod, Crick and Watson, Wally Gilbert, Max Delbruck, Frank Stahl, Francois Jacob, David Baltimore. Of these individuals I have known four in person. I would give a lot to have met Crick and especially Szilard. While at CSHL I learned that James Watson is still alive and intellectually active.

See H. Judson's The Eighth Day of Creation (PDF) for a brilliant but readable history of the golden age of molecular biology.

Yang Wang on Science and Technology in China, Hong Kong Protests, and Corona Virus - Manifold Podcast #34

Yang Wang is Dean of Science at the Hong Kong University of Science and Technology. Professor Wang received his BS degree in mathematics from University of Science and Technology of China in 1983, and his PhD degree from Harvard University in 1990 under the supervision of Fields medalist David Mumford. He served as Chair of the Mathematics department at Michigan State University before joining HKUST.

2:50 - US-China Relations: Has China advanced through the development of human capital or the theft of intellectual property?
16:23 - Academic Culture in China
33:00 - Hong Kong Protests: Economic inequality, housing prices, and outside actors.
1:04:09 - Corona Virus COVID-19: Has the Corona Virus established a new mode of online education in Hong Kong? Yang makes a forecast about the epidemic's trajectory.

Transcript

Yang Wang, Dean of Science at HKUST

Yang Wang (Faculty Profile)


man·i·fold /ˈmanəˌfōld/ many and various.

In mathematics, a manifold is a topological space that locally resembles Euclidean space near each point.

Steve Hsu and Corey Washington have been friends for almost 30 years, and between them hold PhDs in Neuroscience, Philosophy, and Theoretical Physics. Join them for wide ranging and unfiltered conversations with leading writers, scientists, technologists, academics, entrepreneurs, investors, and more.

Steve Hsu is VP for Research and Professor of Theoretical Physics at Michigan State University. He is also a researcher in computational genomics and founder of several Silicon Valley startups, ranging from information security to biotech. Educated at Caltech and Berkeley, he was a Harvard Junior Fellow and held faculty positions at Yale and the University of Oregon before joining MSU.

Corey Washington is Director of Analytics in the Office of Research and Innovation at Michigan State University. He was educated at Amherst College and MIT before receiving a PhD in Philosophy from Stanford and a PhD in a Neuroscience from Columbia. He held faculty positions at the University Washington and the University of Maryland. Prior to MSU, Corey worked as a biotech consultant and is founder of a medical diagnostics startup.

Live Long and Prosper: Genetic Architecture of Complex Traits and Disease Risk Predictors

New paper! Non-coding regions contribute significantly to genetic disease risk -- think twice before you opt for exome sequencing over array genotyping. Also, pleiotropy between common disease risks seems to be weak.

Credit to Soke Yuen Yong for performing essentially all the analysis in this paper.

Genetic Architecture of Complex Traits and Disease Risk Predictors

Soke Yuen Yong, Timothy G. Raben, Louis Lello, Stephen D. H. Hsu

Genomic prediction of complex human traits (e.g., height, cognitive ability, bone density) and disease risks (e.g., breast cancer, diabetes, heart disease, atrial fibrillation) has advanced considerably in recent years. Predictors have been constructed using penalized algorithms that favor sparsity: i.e., which use as few genetic variants as possible. We analyze the specific genetic variants (SNPs) utilized in these predictors, which can vary from dozens to as many as thirty thousand. We find that the fraction of SNPs in or near genic regions varies widely by phenotype. For the majority of disease conditions studied, a large amount of the variance is accounted for by SNPs outside of coding regions. The state of these SNPs cannot be determined from exome-sequencing data. This suggests that exome data alone will miss much of the heritability for these traits - i.e., existing PRS cannot be computed from exome data alone. We also study the fraction of SNPs and of variance that is in common between pairs of predictors. The DNA regions used in disease risk predictors so far constructed seem to be largely disjoint (with a few interesting exceptions), suggesting that individual genetic disease risks are largely uncorrelated. It seems possible in theory for an individual to be a low-risk outlier in all conditions simultaneously.

https://www.biorxiv.org/content/10.1101/2020.02.12.946608v1
doi: https://doi.org/10.1101/2020.02.12.946608

From the conclusions:

III. The DNA regions used in disease risk predictors so far constructed seem to be largely disjoint (with a few interesting exceptions), suggesting that individual genetic disease risks are largely uncorrelated.

Observation III has interesting implications for pleiotropy [63–65]. We found that genetic risks are largely uncorrelated for different conditions. This suggests that there can exist individuals with, e.g., low risk simultaneously in each of multiple conditions, for any essentially any combination of conditions. There is no trade-off required between different disease risks ... One could speculate that a lucky individual with exceptionally low risk across multiple conditions might have an unusually long life expectancy.

Note added: Some clarifying remarks from the comments.

1. We used the output of the UKB variant calling pipeline for the 50k exomes they released -- it is essentially the output data that researchers have available from these exomes. This is discussed in great detail in some of the references as there were some technical issues with the pipeline. SNPs that are not called by this process are (presumably) not determined from the exome reads. Exome sequencing only probes a small fraction of the whole genome, after all.

In any case, we independently analyzed the locations of the SNPs and plenty are outside of coding regions, etc. Referring to the exome process specifically is just to give another "operational" definition of what is in coding vs non-coding regions since the boundaries of these regions are a bit ill-defined in the literature.

2. Plenty of people believe in strong pleiotropy, and are likely surprised by this result. High dimensionality alone is enough to make low pleiotropy plausible, but it *might* have been the case that some special genomic regions play an important role across many diseases. Lots of people with "strong biomedical intuition" told me this would be the case, but apparently not...

There is no way to know until you compare detailed genetic architectures on a disease by disease basis. We are the first to do that.

We don't claim that genetic correlations are close to zero. We just characterize the correlation/overlap between known predictor SNPs for the various diseases. (There is still plenty of heritability not yet discovered for each of the diseases -- need more training data. The predictors will improve a lot in time but these are the most significant SNPs -- i.e. the easiest to discover.)

From our results one can at least put a lower bound on the amount of "risk reduction" (or longevity gain!) that is independently and simultaneously available across various diseases (e.g., if one could make edits freely). It's a lot.