It’s often the case that a single clinic has a great medical treatment that doesn’t spread to the rest of the world.
My favorite example is the Stone Clinic, which for decades has been treating osteoarthritis successfully by regrowing cartilage with stem cells from bone marrow. Extrapolated from their results, if the entire US used the Stone Clinic’s methods, 98% of knee replacement surgeries would be unnecessary. But, for one reason or another, getting widespread adoption is really hard.
Stephen Badylak’s clinic regularly regrows organs, but again, this hasn’t scaled nationwide.
Francis Levi has found that chemotherapy has half the side effects and double the potency if you give it at night, when cancer cells are dividing but healthy cells have a slowed cell cycle. “Chronotherapy” for cancer is pretty much only available at his hospital.
As a patient, you could probably do much better than average by finding the clinic that does that one really good thing and going there.
I would be very interested in reading, say a blog post (or series thereof) exploring why this happens (and, if remotely possible, directing motivated individuals towards ways to support faster adoption of successful treatments).
Doing your own research, though it might take a while. There’s not a centralized database or anything. Skimming Google Scholar for exceptional results on the outcome measure you care about, and then looking up the institution where the study was done, is one way to find it.
Research quality goes way down for stigmatized populations. There’s very little epidemiological evidence about trans people, period. Autism research is frequently “not even wrong” (there are fruit fly and zebrafish behavioral models of autism!)
“Shared common knowledge” within stigmatized communities (e.g. stuff that trans people learn from other trans people) can often outpace the state of the research; it’s higher-variance though, because rumor mills often misinform.
Pain treatment is stupid and bad in a number of ways.
A major reason behind the opioid epidemic is that OxyContin is supposed to give 12 hours of pain relief but it doesn’t actually last that long. Being chronically slightly short of pain medication and suffering withdrawal symptoms is a great way to get addicted.
Opioids don’t actually work on neuropathic pain; they target a totally different pain receptor. Nerve pain is qualitatively different and is sensed by the cannabinoid receptors. And, yet, we don’t have a whole suite of drugs to target the cannabinoid receptors. We just have cannabis, which is illegal in a lot of states. (As a stopgap, cannabidiol is a cannabis compound that helps with chronic pain, is non-psychoactive, and legal.)
“Functional” chronic pain conditions where there’s no visible injury, like fibromyalgia or chronic fatigue syndrome or back pain with a central sensitization component, are very hard to treat and are stigmatized as “all in the patient’s head”, whereas there are a fair number of studies showing that people with these conditions look different hormonally, immunologically, and/or neurologically from healthy people. The standard treatment for CFS for a long time was a graded exercise program which has now been shown to have no evidence of efficacy.
If I were in charge of pain research, I’d study a much wider range of drugs in animals, including new drug classes.
If I had a chronic pain condition, I’d definitely expect self-experimentation and doing my own research to be more useful here than with other diseases, because there’s biases in the medical system (against drug use and disabled people) and a lot of variation between individuals.
Female reproductive health is poorly understood and poorly treated.
Chronic illnesses which are very common (5-21% of American women have fibroids; 15-20% have PCOS; 3% have endometriosis) and cause real problems (infertility and chronic pain) are frequently incurable or nearly so. We don’t know the causes of any of them.. We don’t know how a lot of pregnancy complications (like pre-eclampsia) happen. Anecdotally, women get their reproductive health issues ignored by doctors very often even when symptoms are severe; “period pain” gets written off as trivial even though sometimes it’s severe.
Female sex hormones fluctuate cyclically, of course, but nobody has done studies measuring hormone levels over time and fitting a differential-equation model of how hormones affect each other’s levels. We don’t really know, on the gears level, what happens when you alter hormone levels. Which may be why women get such extremely varied responses to hormonal contraception; “the pill” is one-size-fits-all, but we don’t understand the normal human variation in hormones well enough to target dose to person.
I’m a big fan of normalizing talk about reproductive health in general.
It shouldn’t be considered “TMI” for a man to hear about menstrual or pregnancy or menopause symptoms.
It should be normal to bring up female-specific (or male-specific) side effects of medication, like whether it makes birth control stop working or causes infertility.
It should be normal to talk about miscarriage or infertility, and to have feelings about that.
So, in general, as a techy person looking at biology, you need to be aware that most biomedical researchers are not educated in quantitative stuff. Like, when I worked at a biotech company, I got frequent questions from the bench biologists that amounted to “how do I test statistical significance in this experiment?” where the answer was “do a t-test.”
This means that in any arbitrary field, you’re not necessarily going to find that someone has done the “obvious” applied-math/modeling thing.
Some fields, like genetics or epidemiology or systems biology, have a tradition of collaborating with statisticians or machine-learning people, so the “obvious” stuff will have been tried.
In a lot of fields, you can’t do the “obvious” thing because the data is really expensive to collect. (Measuring how much of each protein is in a sample? Really expensive if you have more than a handful of proteins to test.)
But sometimes, like with hormone dynamics, I’m genuinely puzzled as to why someone couldn’t just do it.
Another example is graph theory for various biochemical networks—metabolic, gene regulatory, etc. Have we looked at connected components, measures of centrality, Rayleigh quotients, etc in attempts to find “master switches”? Surprisingly rarely even when we have the data.
“Systems thinking” is, as far as I can tell, simply “the ability to ask oneself whether one is dealing with a system of differential equations” and it is way rarer than I would guess. I’m not sure why—maybe just the fact that most people have very little math education.
While I still think all of the above is more-or-less true, I’ve since learned it’s not the real reason.
The real reason is that we don’t (yet) have a way to noninvasively measure blood hormone levels as they fluctuate throughout the day. You’d have to keep your study subjects hooked up to an IV, or taking many blood tests a day, continuously for at least a month (because of the menstrual cycle). This is unpleasant and maybe unacceptably risky (infections!) What we need is noninvasive continuous hormone monitoring, which is currently at the prototype stage in a couple of university labs.
Finding somebody who can do the math will be easy once the data exists.
We had a physiology class in my Bioinformatics studies that approached physiology from a systems perspective instead of the usual approach. Unfortunately there was no textbook that our professor could use.
A lot of people would do better, on the margin, to get less end-of-life care than Americans typically do. There are a lot of laws and incentives that prevent doctors from being frank about the fact that the last treatment in a long series of treatments for a terminally ill person doesn’t help that much. The more familiar people are with the medical system, the more likely they are to want to spend their last days at home.
Lifestyle choices in general aren’t usually going to be studied at all, just because there are so many ways to alter your life. It is true that “paleo diets aren’t supported by scientific evidence”—because they haven’t been tested! There’s basically no nutritional research that assigns study participants to <20g carbohydrate diets.
Subtle differences between lifestyle setups are also not going to be studied. Just as the studies didn’t compare “lightboxes for SAD” to “brighter lightboxes for SAD”, you’re usually not going to see a study that compares one type of physical therapy to a slightly different type, one type of standing desk to a slightly different type, one type of exercise routine to a slightly different type, etc. If fine details of execution matter, the research literature often isn’t going to pick up on that.
(This would also explain why I could believe that, say, a study of accupuncture could find it ineffective for pain relief while one particular person going to a particular accupuncturist could genuinely experience a dramatic effect. If some accupuncturists are “real” while others are “fake”, modern medical research isn’t going to be able to distinguish them.)
There are only a few thousand drug candidates being studied in the US. Total. Including in-vitro.
The pipeline is tiny, ultimately because the pharma industry is an oligopoly.
“They haven’t found a drug that does X” really, really does not mean “a motivated and intelligent wealthy donor couldn’t cause someone to find a drug that does X.” It does mean that, short of making some really big political changes, it’s hard to profit on finding such a drug or distribute it at mass scale.
I don’t know where I got that last sentence; that’s clearly bogus. If you knew that a certain drug, target, or research strategy was going to work, of course you could profit off it. That is literally what the biotech industry does.
I’m not sure what past-you meant here, but, one thing you might think is “the amount of hurdles you have to jump through to profit off drugs is ‘hard’, i.e. you (unnecessarily) need to be very well funded and well connected company that can navigate bureaucratic hurdles”, and it’s not that you can’t do it. It’s just, like, “hard”, ya know?
Rare genetic diseases are a classic example of where motivated patients (and parents of patients) can do their own research and make a difference. Sequencing is getting cheap enough for individuals to afford, but hasn’t yet been commoditized that well, so hustling and asking for special favors is useful here. And, of course, if your disease is rare, there isn’t a huge market for treating it.
Matt Might is a famous example of a software engineer who identified and found a treatment for his son’s rare genetic disorder.
In addition to being a skilled software engineer, Matt Might is now a professor, endowed chair, and director of an institute of personalized medicine at University of Alabama’s school of medicine. Before that, he was a professor of computer science at the University of Utah.
I’m just going to list a few fields in medicine that I think stand out as particularly weak.
It’s often the case that a single clinic has a great medical treatment that doesn’t spread to the rest of the world.
My favorite example is the Stone Clinic, which for decades has been treating osteoarthritis successfully by regrowing cartilage with stem cells from bone marrow. Extrapolated from their results, if the entire US used the Stone Clinic’s methods, 98% of knee replacement surgeries would be unnecessary. But, for one reason or another, getting widespread adoption is really hard.
Stephen Badylak’s clinic regularly regrows organs, but again, this hasn’t scaled nationwide.
Francis Levi has found that chemotherapy has half the side effects and double the potency if you give it at night, when cancer cells are dividing but healthy cells have a slowed cell cycle. “Chronotherapy” for cancer is pretty much only available at his hospital.
As a patient, you could probably do much better than average by finding the clinic that does that one really good thing and going there.
I would be very interested in reading, say a blog post (or series thereof) exploring why this happens (and, if remotely possible, directing motivated individuals towards ways to support faster adoption of successful treatments).
Is there a way to find that clinic, given that you know you have condition X?
Doing your own research, though it might take a while. There’s not a centralized database or anything. Skimming Google Scholar for exceptional results on the outcome measure you care about, and then looking up the institution where the study was done, is one way to find it.
Research quality goes way down for stigmatized populations. There’s very little epidemiological evidence about trans people, period. Autism research is frequently “not even wrong” (there are fruit fly and zebrafish behavioral models of autism!)
“Shared common knowledge” within stigmatized communities (e.g. stuff that trans people learn from other trans people) can often outpace the state of the research; it’s higher-variance though, because rumor mills often misinform.
Pain treatment is stupid and bad in a number of ways.
A major reason behind the opioid epidemic is that OxyContin is supposed to give 12 hours of pain relief but it doesn’t actually last that long. Being chronically slightly short of pain medication and suffering withdrawal symptoms is a great way to get addicted.
Opioids don’t actually work on neuropathic pain; they target a totally different pain receptor. Nerve pain is qualitatively different and is sensed by the cannabinoid receptors. And, yet, we don’t have a whole suite of drugs to target the cannabinoid receptors. We just have cannabis, which is illegal in a lot of states. (As a stopgap, cannabidiol is a cannabis compound that helps with chronic pain, is non-psychoactive, and legal.)
“Functional” chronic pain conditions where there’s no visible injury, like fibromyalgia or chronic fatigue syndrome or back pain with a central sensitization component, are very hard to treat and are stigmatized as “all in the patient’s head”, whereas there are a fair number of studies showing that people with these conditions look different hormonally, immunologically, and/or neurologically from healthy people. The standard treatment for CFS for a long time was a graded exercise program which has now been shown to have no evidence of efficacy.
If I were in charge of pain research, I’d study a much wider range of drugs in animals, including new drug classes.
If I had a chronic pain condition, I’d definitely expect self-experimentation and doing my own research to be more useful here than with other diseases, because there’s biases in the medical system (against drug use and disabled people) and a lot of variation between individuals.
Female reproductive health is poorly understood and poorly treated.
Chronic illnesses which are very common (5-21% of American women have fibroids; 15-20% have PCOS; 3% have endometriosis) and cause real problems (infertility and chronic pain) are frequently incurable or nearly so. We don’t know the causes of any of them.. We don’t know how a lot of pregnancy complications (like pre-eclampsia) happen. Anecdotally, women get their reproductive health issues ignored by doctors very often even when symptoms are severe; “period pain” gets written off as trivial even though sometimes it’s severe.
Female sex hormones fluctuate cyclically, of course, but nobody has done studies measuring hormone levels over time and fitting a differential-equation model of how hormones affect each other’s levels. We don’t really know, on the gears level, what happens when you alter hormone levels. Which may be why women get such extremely varied responses to hormonal contraception; “the pill” is one-size-fits-all, but we don’t understand the normal human variation in hormones well enough to target dose to person.
I’m a big fan of normalizing talk about reproductive health in general.
It shouldn’t be considered “TMI” for a man to hear about menstrual or pregnancy or menopause symptoms.
It should be normal to bring up female-specific (or male-specific) side effects of medication, like whether it makes birth control stop working or causes infertility.
It should be normal to talk about miscarriage or infertility, and to have feelings about that.
>nobody has done studies measuring hormone levels over time and fitting a differential-equation model of how hormones affect each other’s levels
What in the everloving fuck? That really seems like the first thing you should do. Has that at least been done for the shared hormones?
So, in general, as a techy person looking at biology, you need to be aware that most biomedical researchers are not educated in quantitative stuff. Like, when I worked at a biotech company, I got frequent questions from the bench biologists that amounted to “how do I test statistical significance in this experiment?” where the answer was “do a t-test.”
This means that in any arbitrary field, you’re not necessarily going to find that someone has done the “obvious” applied-math/modeling thing.
Some fields, like genetics or epidemiology or systems biology, have a tradition of collaborating with statisticians or machine-learning people, so the “obvious” stuff will have been tried.
In a lot of fields, you can’t do the “obvious” thing because the data is really expensive to collect. (Measuring how much of each protein is in a sample? Really expensive if you have more than a handful of proteins to test.)
But sometimes, like with hormone dynamics, I’m genuinely puzzled as to why someone couldn’t just do it.
Another example is graph theory for various biochemical networks—metabolic, gene regulatory, etc. Have we looked at connected components, measures of centrality, Rayleigh quotients, etc in attempts to find “master switches”? Surprisingly rarely even when we have the data.
“Systems thinking” is, as far as I can tell, simply “the ability to ask oneself whether one is dealing with a system of differential equations” and it is way rarer than I would guess. I’m not sure why—maybe just the fact that most people have very little math education.
While I still think all of the above is more-or-less true, I’ve since learned it’s not the real reason.
The real reason is that we don’t (yet) have a way to noninvasively measure blood hormone levels as they fluctuate throughout the day. You’d have to keep your study subjects hooked up to an IV, or taking many blood tests a day, continuously for at least a month (because of the menstrual cycle). This is unpleasant and maybe unacceptably risky (infections!) What we need is noninvasive continuous hormone monitoring, which is currently at the prototype stage in a couple of university labs.
Finding somebody who can do the math will be easy once the data exists.
We had a physiology class in my Bioinformatics studies that approached physiology from a systems perspective instead of the usual approach. Unfortunately there was no textbook that our professor could use.
Nope! I went looking! Not there!
A lot of people would do better, on the margin, to get less end-of-life care than Americans typically do. There are a lot of laws and incentives that prevent doctors from being frank about the fact that the last treatment in a long series of treatments for a terminally ill person doesn’t help that much. The more familiar people are with the medical system, the more likely they are to want to spend their last days at home.
Lifestyle choices in general aren’t usually going to be studied at all, just because there are so many ways to alter your life. It is true that “paleo diets aren’t supported by scientific evidence”—because they haven’t been tested! There’s basically no nutritional research that assigns study participants to <20g carbohydrate diets.
Subtle differences between lifestyle setups are also not going to be studied. Just as the studies didn’t compare “lightboxes for SAD” to “brighter lightboxes for SAD”, you’re usually not going to see a study that compares one type of physical therapy to a slightly different type, one type of standing desk to a slightly different type, one type of exercise routine to a slightly different type, etc. If fine details of execution matter, the research literature often isn’t going to pick up on that.
(This would also explain why I could believe that, say, a study of accupuncture could find it ineffective for pain relief while one particular person going to a particular accupuncturist could genuinely experience a dramatic effect. If some accupuncturists are “real” while others are “fake”, modern medical research isn’t going to be able to distinguish them.)
There are only a few thousand drug candidates being studied in the US. Total. Including in-vitro.
The pipeline is tiny, ultimately because the pharma industry is an oligopoly.
“They haven’t found a drug that does X” really, really does not mean “a motivated and intelligent wealthy donor couldn’t cause someone to find a drug that does X.” It does mean that, short of making some really big political changes, it’s hard to profit on finding such a drug or distribute it at mass scale.
I don’t know where I got that last sentence; that’s clearly bogus. If you knew that a certain drug, target, or research strategy was going to work, of course you could profit off it. That is literally what the biotech industry does.
I’m not sure what past-you meant here, but, one thing you might think is “the amount of hurdles you have to jump through to profit off drugs is ‘hard’, i.e. you (unnecessarily) need to be very well funded and well connected company that can navigate bureaucratic hurdles”, and it’s not that you can’t do it. It’s just, like, “hard”, ya know?
Rare genetic diseases are a classic example of where motivated patients (and parents of patients) can do their own research and make a difference. Sequencing is getting cheap enough for individuals to afford, but hasn’t yet been commoditized that well, so hustling and asking for special favors is useful here. And, of course, if your disease is rare, there isn’t a huge market for treating it.
Matt Might is a famous example of a software engineer who identified and found a treatment for his son’s rare genetic disorder.
In addition to being a skilled software engineer, Matt Might is now a professor, endowed chair, and director of an institute of personalized medicine at University of Alabama’s school of medicine. Before that, he was a professor of computer science at the University of Utah.