Balaji Srinivasan's AMA on China's guidance to the diagnosis and treatment of COVID-19

So to just go ahead and get started, I want to thank everybody for coming out today. I'm really excited for a call speaking with Dr. Balaji Srinivasan. So one of the reasons I've been really excited about this is I'm a medical student on the east coast who's actually scheduled to start clinical rotations yesterday. And those have obviously been pushed back. And so the flow of information about this viral disease that's spreading around the world has been fascinating in that I've learned a lot more about it from people like Walaji on Twitter than I have from sort of the traditional sources of medical information. So to me, I think this is a really great opportunity to learn from someone who's become an expert in the topic over the past couple of months and learn about how information spreads within the global health ecosystem and how we can help improve it. So to get started, Balaji, do you mind giving a quick introduction about yourself?

Yeah, sure. So, you know, my name is Balaji. I am Stanford Lifer, BSMS, PhD, Electrical Engineering, MS, Chemical Engineering. I did my PhD work in clinical and microbial genomics. Basically, like genetic circuits of microbes, like reconstructing the maps of how proteins and RNAs and other kinds of things all talk to each other within the cell. That's at the intersection of like stats and genomics and molecular biology and electrical engineering and so on. After that, I taught cell CS and stats as a lecturer at Stanford for a few years. And then rather than going and becoming a professor, I went and started a genomics company, Molecular diagnostics lab, actually, which we scaled and ended up selling for 375 million. Then, you know, I got into crypto. After that I was basically early investor in Bitcoin and, you know, Ethereum, zcash. That's how I got to know Brian Armstrong. Was doing research and also was a general partner at Anderson, Harvard's multibillion dollar venture capital firm. Helped set up what became our bio and crypto funds. Early investor in Soylent, Superhuman, Lambda School, Cameo, a bunch of other companies like that. And Most recently sold Earn.com to Coinbase from the CTO of Coinbase. And then going to start a new thing this year. And then in January, I kind of got involved with this COVID 19 thing back when it was still in COVID 19. And that brings us to the present day.

Yeah, thank you for that. Clearly you kind of have a lot of experience that you can bring into this conversation. So I guess the first question, just to kind of frame things is when did COVID 19 come to your attention? And how long after that were you like, wow, this is something that's going to kind of change the whole world?

Well, so, you know, I had actually paid a lot of attention to the ebola thing in 2014, which seemed, you know, is making its way into the US and there were breakdowns in process. And, you know, there's a. There's a woman in, I believe, Texas that got infected. It was a nurse. And so, you know, I saw this, you know, new virus in January. But what really got my attention was because it was kind of tracking in my peripheral vision a thousand other things. What really got my attention was what got, obviously a lot of people's attention. But it was the lockdown of Wuhan and Hubei province, I believe, January 23rd. And that was an unfavable signal because that's just not the kind of thing that China had typically been known for doing, since their entire sort of understanding with the population is the population may not have a political say, but the party delivers economic growth. And lockdown was something that seemed sacrificed economic growth. And so it had to be extremely serious for them to do something like that. And so that got me really interested in what was going on, because that was not. That was not a fakeable signal.

Right.

Unless they were making that up. That was absolutely unprecedented. So I started reading into literature and looking at stuff, and I got more and more, you know, concerned, and that. That's basically how it came to my attention.

Yeah, it makes a lot of sense. I mean, shutting a whole city down. I think we've all probably seen the drone footage. You know, it's. It's extremely eerie. Reminds me of a zombie movie.

Yeah.

So that. That makes a lot of sense to me. And I think, you know, I was very excited when you said you wanted to share this paper, because I think there's a lot of insight in it. I was looking at your Twitter last night. Night. And I think over the past couple of days, there's been a lot of talk about the concept of universal basic masks. So being able to kind of prevent transmission from one person to the next by essentially stopping the respiratory system from people inhaling the virus. One of the most interesting things about this to me is how some of the centralized health organizations like the cdc, the WHO had put out recommendations basically saying masks don't help stop the spread of the virus. Within this document, there's actually a section about transmission where I'll quote it really quick. It says, transmission of the virus happens mainly through respiratory Droplets and close contact. There is a possibility of aerosol transmission in a relatively closed environment for a long exposure of time. High concentrations of aerosol. So they're basically saying they know it occurs through respiratory droplets and maybe occurs through aerosol transmission. The medium article you shared essentially said that a surgical mask has the ability to stop these large particles from entering a new person's respiratory system and spreading the virus. I guess to me, that's really impressive because the Chinese government has essentially put out information that's not necessarily contradictory to the cdc, but maybe has a little bit more resolution to it. So I'm curious if there's anything else within this document that you were like, wow, this needs to be spread among health care workers in the United States and Europe in order to kind of stay a step ahead of maybe the centralized organization's perspective on this disease.

Well, so, I mean, there's a bunch of things. Well, first is, I mean, you know, one of the things I've talked about on Twitter is the degree of what I call official misinformation on this has been ludicrous. Because, you know, the problem is that if it's just some random person saying something on Twitter, you can, you can argue with it more easily. But, you know, just. Just a partial list. You know, folks are saying it's just the flu, the travel bans are overreactions, and only people from Wuhan were at risk. Avoiding handshakes is paranoid. The virus is contained, tests are available, and then masks don't help. Which, you know, I tweeted this yesterday, you know, but basically, Surgeon general's warning. Don't listen to the surgeon general's warning about masks because they do help. Everybody should wear a mask. It's the simplest thing we can do upstream to, you know, limit contagion. And everyone in China was wearing masks. And yeah, even if there's an N95 shortage, you can. You can improvise a mask and wear that. So that is certainly at variance with where CDC and WHO are. But I think they're going to switch. Washington Post reported that they're considering a change in guidelines. And it's such an obvious thing to do, and the literature supports it. It's really silly that they're not. We asked what other things are there besides that that folks aren't doing? I mean, there's a ton. I mean, the most important is just like reading the guidelines. There's lots of individual clinical things, but macro things. For example, isolation of people immediately.

Right.

They had separate hospitals for Covid and non COVID patients. And if you have Covid, you're not kept with your family. You're isolated, you know, separately. And they have a. They have. Well, they have recommended treatments here. You know, they have hydroxychloroquine and they have. Or rather chloroquine is in here. They have convalescent plasma. They have, you know, like. Like all the stuff that folks are figuring out the. The Chinese, you know, I sent this to a friend of mine. He's like, this document is from the future. You know, it's like, oh, this is after we nailed the disease. And, you know, we had the guidelines written by, like, you know, 2022 or 2023. This is what you have. Something like that, aside from the traditional trans medicine part. So, yeah. So high level, you know, another big thing that's an obvious thing that's not a very narrow clinical guideline is separate hospitals and isolate people away from their family. So it's as important as masks?

Yeah, absolutely. I think the one thing I find really interesting is the phrase that you said, the Surgeon General warns not to listen to the Surgeon General. Which is crazy. Well, no, no, no.

I'm just. Well, so what I'm saying is that was a play on words. You know, on a pack of cigarettes, you'll probably see this thing, Surgeon General's warning. Tobacco can cause this and that, right? And there's a tweet by the Surgeon General of the United States as of today, which said, hey, people, don't wear masks, et cetera. And I retweeted that sarcastically, surgeon General's warning. Don't listen to the Surgeon General's warning.

Anyway, whatever, go ahead. No, no, I think it's actually really apt and kind of is a useful analogy for people who are in the healthcare industry today because. So I'm a medical student, and we recently just had a presentation on COVID 19 and kind of how to protect yourself while in the hospital. And the content of that presentation came from, like, the CDC and some of these more, I guess, slow to move centralized entities. So I'm curious, like, if you are a healthcare worker today in the United States and you're kind of getting the understanding that maybe like, you know, the Surgeon General warns not to listen to the Surgeon General, who should we listen to? Where should we turn to try and find information that's going to be the most effective at protecting our patients and then ourselves?

Well, so here is the short answer is people who are capable of reading the scientific and biomedical literature on this, and you know that you can Limit that if you want to, to folks who have an academic background in these topics. I can understand why people would want to do that, but let me, let me dwell on that point just for a second. There's what I call pre headline people and post headline people. Post headline people are those who pretty much won't believe something unless some official institution has already signed off. Pre headline people are folks like journalists, like academic researchers, or like entrepreneurs, venture capitalists who make the news.

Okay.

And so, like, you know, a journalist doesn't need to read something in a headline. In fact, if they've read in a headline, they've been scooped. So they're a pre headline person. Obviously. Same with a researcher. A researcher might discover a true fact. You don't need some central authority to sign off on that true fact that the fact that they've discovered it, you know, like they actually want to write it up and publish it first.

Right.

Same with an entrepreneur or venture capitalist. So that's actually how the literature is written. There's somebody who is not the surgeon general who is actually doing the research that the surgeon General is quoting.

Right.

And that's the group of pre headline people. Okay. So first, that's a group. Second is that group is actually fairly large. It's essentially the group of people who have what I call, Right, access to medical guidelines. Okay. Most doctors have essentially just read access to medical guidelines. They are practitioners. They do not have license really to rewrite the textbook. Okay, we can talk about why that is. Partly it's due to liability concerns. Partly it's due to, you know, like medicine being turned into something where you get a paycheck at a hospital as opposed to being kind of an independent, swashbuckling kind of person. But medicine has become more centralized, and so there's a lot of kind of nodes that are downstream that are just executing on commands from these central institutions like New England Rural Medicine, the Sheriff General, cdc. Those guidelines come down and then folks just kind of blindly execute on them. That's most MDs that are out of the field. But there is a small group, relatively small, but not tiny, of, I guess, I don't know the exact number, but certainly more than hundreds, probably in the range of thousands of academics who have license to rewrite the textbook. Okay. And you know, nephrologists and, you know, diagnostics people and so on can rewrite those things. And those folks can go toe to toe with a bad recommendation that's gone put out there.

Okay.

And the good thing is that's actually a Fairly large group and many of them are actually spitting mad right now over, you know, some of the bad guidelines that are being put out there. And so that's where you're seeing UW and UCSD doing things, UW doing its own testing, UCSD is building its own ventilators and so on. So folks are willing to get out there and those would be the folks you'd go to. And it was a long answer, but maybe that was clear.

It's a complicated thing, you know, trying to curate your own list of experts that you can believe in any moment, I guess. So if you're a pre headline person, who has that right access to the medical guidelines, how do you share that information in real time? Because things change almost on like a daily basis. So if I'm a clinician, how do I read the content that those people are producing in order to stay as up to date as possible?

Well, I mean the thing is that that's in theory, that's why they're supposed to pay doctors big bucks, is they're supposed to be, you know, or at least I think the general public's impression of doctors is as folks who make a lot, a lot of independent decisions who are balancing X and Y and Z and W and you know, they're reading the literature and looking at your chart and you know, stroking their chin and their house MD ing it, if you remember that TV series. But, but most of them don't have that, you know, discretion really, but they're starting to now. So my short answer is usually doctors in the field just have to go by guidelines because they have liability issues if they don't go by guidelines. But now there's actually a Bloomberg article that I tweeted about iterative medicine coming back, which is very important. And now what's happening is those doctors are talking to other doctors and they're sharing what works, then they're going and using that. And you know, that is like getting them better results with patients than waiting for a paper.

Yeah, that makes a lot of sense. I know personally just a couple of people in my area where I agree with you. I think a lot of physicians, basically, they have information dictated down to them in a top down sort of way. But with this specific scenario, as the information is coming so quickly, you almost have to kind of gather it in a more grassroots manner. Despite that though, I know doctors who have tried to do simple things like essentially take the temperature of everyone who comes into their hospital and there are administrative roadblocks that they face because of this. So it's an interesting scenario in how you can actually implement those in an effective way to act on the information that you're talking about.

Well, you know, we have something unfortunate that may happen, which is the only institutions that survive may be those that move at the velocity of the virus. So what that means is anything that is not absolutely essential to treating patients, it's essentially like, I mean, you take somebody really out of shape and then you just make them work out incredibly hard. You know, this is taking an incredibly flabby and out of shape process that's suffused with red tape at every stage. That's optimized for optics rather than substance. And it's making it generate substance and making it generate startup like results where you have to put it together with duct tape and baling wire, but you must ship it now, yesterday.

Right?

So, so speed, you've never had something before in the medical system where like everybody in the entire country was focused on it. Number one, okay. Number two, speed was undeniably important. Number three, it was like lethal. Number four is like international and infectious. It combines a bunch of different things at the same time. Like, AIDS may be the closest thing, but it didn't. AIDS doesn't have the spectacular progression of this disease in the sense that, you know, you can get infected and die within weeks. And that's actually a fairly common occurrence. I mean, it's not like one in a million people that has that happen to them. And it can happen to relatively young people, you know, in their, in their 40s and 50s and some, you know, some of their 20s. So because of that, all of those things, like the administrative thing you're talking about, if that continues, the bodies literally will pile up in that hospital. And that will actually be unfortunately the feedback signal that people use to just shove those things out of the way and actually do what is necessary. Hopefully.

Yeah, it makes sense. I mean, that's definitely something you can't really ignore. So bringing it back to the paper a little bit, I think this is an interesting, because I really believe that this paper has a lot of value in it and a lot of the information, as you said, kind of feels like it's coming from the future. Two things specifically, which really kind of like got my attention. One, in the treatment section, there is a significant portion that's dedicated to traditional Chinese medicine. So I'm curious what you think about that.

So I mean, you know, the, like the continuum between a food and so traditional Chinese medicine, it's basically like a whole section of the for those who haven't read the paper, there's a. There's a whole section, or the guidelines, rather, there's a whole section devoted to essentially, like, these potions that you're supposed to drink. And, you know, they refer to it in terms. It's really interesting that that is in the official national guidelines. You know, they put it. They give it pride of place alongside all the allopathic stuff. You know, in one section, they're talking about the, like, the dimensions of the virus and how many microns it is or, you know, nanometers and. And then, you know, just in a later section, they're talking about, okay, if you have like a. I mean, pronounce this wrong, but like a Quinn deficiency, you know, what do you do?

Right.

And. And that's, like, kind of amazing to us. Right. But what I think it probably is, just given, you know, how effective they've been in the rest of the document, it's possible that it's purely psychosomatic. It's also possible if there's active ingredients in those potions that they're pulling together and they just don't know what they are yet.

Right.

Because, you know, like, the continuum between a food and drug, it is a continuum. For example, like, lovastatin is present in red yeast rice. And, you know, the FDA actually made people remove the lovastatin from red yeast rice before it could be important because otherwise people would be able to get access to lovastatin within this food. Right. So there's many, many drugs that are basically food like, or vice versa. And I wouldn't be surprised if there were some powerful drugs or combination drugs, you know, in these potions that they're cooking up in the traditional Chinese medicine section.

Yeah, I actually dug into it a little bit, and one of the recommendations that they have for, like, mild to moderate cases of lung congestion essentially is a tea that has a plant called ephedra in it. And one of the ingredients is ephedrine, which essentially is the same pharmaceutical mechanism as, like, a beta agonist that you would use to kind of open up lungs during asthma. So it just struck me that, as you said, the TCM recommendations were right next to kind of the more allopathic recommendations in such an official document.

Yeah, no, I mean, like, I think, you know, I think we're still actually very primitive in our understanding of the body. You know, like, I used to be extremely. You know, I certainly wouldn't call myself a bull on it, but I used to be extremely skeptical of stuff like Acupuncture and, you know, then somebody actually put needles in my arm and they said, oh, this meridian is connected to this other thing and I could like wiggle my hand and this thing, the needle would actually move in your leg or whatever.

Right.

There's actually something that was happening there. And can I say that it's effective? No. I don't know. But I do think that it's worth investigation, I think it's worth study. And I think that with diagnostic grade wearables or something that'll give you actually a stream of telemetry out of your body, I think that would actually be very useful for actually starting to calibrate what foods are good for you. Which of these treatments actually work in a completely different way than normal? You just have like 10,000 streams of data coming out of you over time, which a Fitbit and an Oura ring and so on are all like V1. And then from that you can look at what your stats are looking like, what transcription factors are binding now and what genes are getting turned on and what does your metabolomic profile look like and so on. As you take these treatments as well as as you take foods, and you can start to do systems analysis, figure out what the stimulus response is. Right now we're in an incredibly primitive state in medicine because you have no baseline. Where the baseline you get is kind of a primitive lab, you know, where you're doing it, I don't know, every 12 months or something like that. That's not a true, true kind of time series. You want it kind of continuously. Once we have that, I think we'll be able to benchmark traditional Chinese medicine, acupuncture, you know, different kinds of drugs, different kinds of foods, all these things on your body, and do so in a coherent, you know, framework.

I think you bring up a really good point there in that. Sharing of data, I think, is a really critical way to try and cope with a disease that moves this quickly. One of my biggest frustrations with medicine is that electronic health records are really hard to share based on essentially privacy laws. And if there was a way to somehow use big data analytics on all the health information that was being produced in a country at any one moment, I think there would be a lot more insights into what's actually working and not working, as opposed to kind of small trials that end up getting published in preprints. I'm curious if you have any thoughts there when it comes to clinical data combined.

Absolutely, I have thoughts and it's a great prompt, essentially. Tyler Cohen actually had a good post on this where he was observing the degree of heterogeneity or apparent heterogeneity in things like case fatality rates or, you know, severity of cases, hospitalization rates. It's like, why is Italy so bad? But you know, why does it not seem to be going up in San Francisco or you know, so on. And really what you want is you want a table, okay, where rows are patients and columns are features and you've got, you know, the, the cases, you know, so each case is a row, right? And you might have the first field, but you like, like a, you know, binary blob from which you extract other fields like their height and weight and gender and age and you know, did they smoke? And so on and so forth. You have all these, all these variables going across, like just a gigantic spreadsheet with rows being patients and common features. Then you have the, you know, the time series of the progression of that patient. You know, on day one, what happened, on day two, what happened. And so you have this effectively this fusion, this gigantic table. If we had that table, okay, we could, we could tell things. We could say instantly look at it and we could say, oh, well, the reason that Italy has a higher death rate is that these patients have, you know, they're older on average or they all have, for example, this genetic variant or you know, if they, you know, they all came in at this XY location. So if we had that gigantic table, there's a ton that we could say and we probably would understand the heterogeneity more. We don't. And the reason we don't is we don't have any easy way for folks in this pandemic to just click a button and upload know and download case data. Now what I'm talking about by the way, is incredibly complicated, okay? All the EHR people or you know, like not, not the bioinformatics but health informatics people would immediately say, well, all these countries have different systems and their coding is different and you know, what one one area codes as severe. The other guys may not code as severe. They might code it as mild and so on. So for example, like diagnosed with diabetes, it's, it's not always a 01. It's, it can be continuous and different thresholds you can often set. So that's all true, but unless we resolve that, we're giving up tons of data and you're kind of reasoning through a pinhole camera, right, because you're seeing something in the literature, but that what's in the literature is an aggregation across, you know, some columns of this theoretical full data set.

Right.

And if you're the full data set, you can do analyses that none of the individual authors thought about. I've seen the power of this in genomics, where it actually is relatively easy to take studies that have been done by different groups and stack them together. Okay. And the reason you can stack them is the human genome is roughly constant, so the same variants appear in each one. So you can stack a bunch of different studies together and therefore get signal. And so we want to get or basically stackable data sets so that Italy's data plus Spain's data plus China's data plus the US's data and so on. There's a strong incentive for people to share because if you share, you get more signal. So you give to get. And same with a patient, you know, a patient who shares and does give up basically. Maybe some privacy. I'll come back to that point. Let's just say they give up some privacy by sharing in return is able to benefit from the statistical aggregation where they can see how typical or atypical that case is relative to others. I mentioned the privacy component. I think there's a ton of work that can be done here in terms of privacy preserving navigations. There's Google Private and Trillion compute that recently came out. There's secure multi party computation. There's a lot of stuff you can do where you don't necessarily need to have your medical history just out there, you know, on, you know, not on the Internet, but for 10,000 doctors or whatever to look at. There's ways you can do privacy preserving stuff potentially. But I think this is a useful mental concept to have, which is the big table with Rose patients and Tom's features. And from that we could figure out a ton.

Yeah, I do hope that one of the silver linings of this whole scenario is that there's pressure to accelerate that type of development within the healthcare system, bringing it back to the paper. And this is more guidelines than like a traditional scientific paper. But one thing which I was curious about is that the recommendations are really good, but there are no citations pointing to studies to basically allow like a critical eye to maybe look at things a little deeper. Does China have a big table like that and is that what they're using to essentially make these recommendations? I'm curious if you have any insight into where they come from.

Yeah, so the thing is, you know, I tweet about this a bit, but basically, you know, this is. I think where so called evidence based medicine starts to, to, should I say fall down? That's not exactly right, but here's the thing. So evidence based medicine, people will give you different definitions, but evidence based medicine is essentially, it's the idea that every treatment decision must be made on the basis of evidence. And the best forms of evidence are peer reviewed studies and meta analyses and randomized controlled trials and so on. And otherwise the idea is that you're basing medical judgments not on rigorous evidence, but on hearsay. And it's kind of like it's sort of set up in such a way to be a dis. Like, oh, I practice evidence based medicine. What do you do? Oh, you just like heard a rumor from some doctor and you're prescribing it that way. That's pretty fly by night, man. You know, and that's, that's kind of how it's constructed, right? Like evidence based medicine is supposed to be the rational form of medicine and everything else is just, you know, like people playing with sticks or whatever, right? Or just doing it by the seat of their pants.