4 ms·

It’s become increasingly clear that genetics play a huge role in which psychiatric drugs are effective and which ones are not. Genetic testing for psych med eff

It’s become increasingly clear that genetics play a huge role in which psychiatric drugs are effective and which ones are not. Genetic testing for psych med effectiveness should become standard practice given how in some people, SSRIs (often the first line of defense in psych meds) can cause extended periods of psychosis. Funny thing about psychosis — when you’re psychotic, your erratic and extreme behavior doesn’t feel abnormal. You don’t feel like anything has changed. This happened to me on Prozac — the most commonly prescribed antidepressant — and it took me 3 years to put my life back together after all the damage I did to my personal relationships during that period. The anxiety I was dealing with was crippling, but it didn’t dismantle my life the way the side effects of that drug did. The treatment was absolutely worse than the disease. If they had done a genetic test first, they would have seen a red flag and avoided SSRIs all together. This testing exists today, but it’s only used in patients who show strong resistance to pharmacological intervention. It should be standard.

SmokeGS 7y ago
I feel like we are so close to wirelessly monitoring all brain waves (next 50 years) we will probably be able to, in real time, recognize and diagnose abnormal brain patterns. I think it would be promising if we get this sensor tech up and then perhaps deploy nanotech to interfere with maladaptive brain patterns. The ethics of such a device is considerably sticky. TED talk by Jack Gallant https://www.youtube.com/watch?v=Ecvv-EvOj8M https://www.youtube.com/watch?v=Ecvv-EvOj8M recreating vision: https://www.youtube.com/watch?v=nsjDnYxJ0bo https://www.youtube.com/watch?v=nsjDnYxJ0bo another ted talk by John-Dylan Haynes https://www.youtube.com/watch?v=mMDuakmEEV4 https://www.youtube.com/watch?v=mMDuakmEEV4 paraphrased: "wireless non-contact mind reading tech may be cheap and mass producible within the next 50 years." the only rate limiting step with his method of decoding thoughts is that it takes time under a high energy device to pair associations in the brain. Think of how much screen time we currently use. If we could pair a wireless 'headset' like headphones while looking at pictures on a screen it would be possible to build a brain bank of children->tenenagers. Currently we just need a lot of processing power and reserved bandwidth on the internet to do some cool stuff.
- SubiculumCode 7y ago
  In my honest opinion as a neuroscientist, we are not that close. We have only an obtuse understanding of how the brain works. Most 'mind reading' demonstrations, though cool, occur within highly constrained circumstances and training and test data, and even then perform just a bit over chance. For context, we barely achieve good accuracy classifying ASD vs typical development, ADHD vs typical development using neuroimaging, much less the much harder problem of classifying and decoding the contents of our thoughts. Slightly more impressive results have been achieved in studies collecting data from actual brain implants found in epilepsy patients, but even then, I do not see this as something that can be a reliable technology in a long long time.
  - SmokeGS 7y ago
    I appreciate someone in the field taking the time to respond. I think we have the ability to scale our knowledge with machine learning and mass monitoring. Again i recognize that 50 years is extremely optimistic. Ethics, money, and politics never agree.
    - feanaro 7y ago
      The poster before you is presumably talking about physical limitations of the current measurement methods which leads to very high noise. I doubt machine learning can overcome that.
      - SubiculumCode 7y ago
        I think there are several issues. 1. Our understanding of the brain is poor, and I think that will be necessary to have good machine decoding. In principal, a machine learning algorithm does not need to know how the brain works to make an accurate decoding prediction, but I suspect understanding will be needed to help construct and constrain machine learning models learn to deal with the myriad of mental states, intentions, emotions, etc that can occur. While we can train classifiers better than chance to identify whether one or another item is observed, remembered, etc, these training are quite constrained to a type of information or class. I imagine those classifiers are hopeless when you start performing another activity or mental state, and the machine would need to recognize the new state and apply different decoders. Second, yes there is too much noise in most available methods (EG, fMRI, MEG) and the information is not dense enough (spatial, temporal resolution) and not specific enough to the type of activity (excitation, inhibition, chemical diffusion, etc). Electrodes have done cool things, but I suspect that even millions of wires would not provide enough information and types of information to provide a full fledged brain interface. However, I'll admit, tech is moving so fast...
hurrdurr2 7y ago
"Treatment is worse than the disease". This is also true in cancer treatment, where many drugs have measured effects on reducing tumor size and yet not increasing survival times...yet the side effects make the patients' quality of life significantly worse.
- polyfractal 7y ago
  I don't think this is fair for several reasons. First, cancer treatment drugs are tested heavily in-vitro first, where the only significant metric is reduction of tumor mass. Clinical trials for cancer treatments are different from other drugs, as you don't generally do double-blind trials (would you want to get a placebo if you had a terminal cancer? No, probably not). So many cancer treatments go into clinical trials with only the evidence that it shrinks tumors, and tends to kill tumors faster than it kills the rest of your body. Side effects are expected to be severe since most cancer treatments are quite literally killing any cells that divide... tumors just divide faster and thus get hit harder. The end result is cancer treatment often has very poor quality of life. Clinical researchers _do_ perform analysis to see if survival increases relative to other treatments. But often you don't know until you've already tried it on a cohort of patients (with their consent, obviously). There is push-back in the medical field in areas for what you suggest. For example, colonoscopies are becoming widely recognized as causing more problems than they solve. The complications they can induce tend to lead to a poorer patient outcome than missing colon cancer because you didn't screen (e.g. survivability is not affected by finding/treating colon cancer early). I just don't think it's fair to paint cancer treatments with the same brush strokes, since dealing with terminal diseases is a totally different ball game. Edit: I should say, there are some parts of cancer treatment which is moving towards the dont-treat opton. For example, there is growing evidence that treating prostate cancer is not necessary, since you'll probably die of something else long before prostate cancer kills you (it is very slow growing on average)
  - cannonedhamster 7y ago
    The other problem with cancer is that there are different types of cancer cells. There's the fast dividing and there's also the cancer stem cells which can lay dormant for years without dividing. This is the reason for most cancer returning. There's a promising new tech that hopefully pans out that activates these cells so standard chemo can attack them. Regarding clinical trials, they are very selective. You have to for a very specific criteria and your cancer needs to be measurable by a CT scanner typically to prove efficacy. Add to that the typical cancer patient is either a child or an older person and you're bound to have poorer results.
epmaybe 7y ago
Good luck figuring out how to sustainably pay for personalized precision medicine across the board.
- nitrogen 7y ago
  Simple: keep doing more of it until it's cheap.
ppod 7y ago
>This testing exists today, but it’s only used in patients who show strong resistance to pharmacological intervention. It should be standard. Is it possible to do this kind of thing oneself with 23andme style SNP tests and a google scholar search, or does it involve more than that?
- derekja 7y ago
  only slightly beyond that, 23andme doesn't cover all of the relevant SNPs but it does cover the ones that allow you to infer that you are less likely to respond to p-glycoprotein based drugs like citalopram and paroxetine. You have to export your 23andme raw data and use something like promethease to reprocess the data.
- JamesBarney 7y ago
  No, the poster is seriously over estimating the effectiveness of genetics testing.
DuskStar 7y ago
Isn't this mostly from companies like GeneSight? Considering that they relied on 5-HTTLPR and HTR2A, which both had enormous edifices of completely useless (and now debunked) scholarship behind them, I wouldn't bet on those tests' effectiveness. (For reference, GeneSight's test relied on 7 genes. The five not mentioned deal with liver enzymes) https://slatestarcodex.com/2019/05/07/5-httlpr-a-pointed-review/ https://slatestarcodex.com/2019/05/07/5-httlpr-a-pointed-rev...

SmokeGS 7y ago

I feel like we are so close to wirelessly monitoring all brain waves (next 50 years) we will probably be able to, in real time, recognize and diagnose abnormal brain patterns. I think it would be promising if we get this sensor tech up and then perhaps deploy nanotech to interfere with maladaptive brain patterns. The ethics of such a device is considerably sticky. TED talk by Jack Gallant https://www.youtube.com/watch?v=Ecvv-EvOj8M https://www.youtube.com/watch?v=Ecvv-EvOj8M recreating vision: https://www.youtube.com/watch?v=nsjDnYxJ0bo https://www.youtube.com/watch?v=nsjDnYxJ0bo another ted talk by John-Dylan Haynes https://www.youtube.com/watch?v=mMDuakmEEV4 https://www.youtube.com/watch?v=mMDuakmEEV4 paraphrased: "wireless non-contact mind reading tech may be cheap and mass producible within the next 50 years." the only rate limiting step with his method of decoding thoughts is that it takes time under a high energy device to pair associations in the brain. Think of how much screen time we currently use. If we could pair a wireless 'headset' like headphones while looking at pictures on a screen it would be possible to build a brain bank of children->tenenagers. Currently we just need a lot of processing power and reserved bandwidth on the internet to do some cool stuff.

SubiculumCode 7y ago
In my honest opinion as a neuroscientist, we are not that close. We have only an obtuse understanding of how the brain works. Most 'mind reading' demonstrations, though cool, occur within highly constrained circumstances and training and test data, and even then perform just a bit over chance. For context, we barely achieve good accuracy classifying ASD vs typical development, ADHD vs typical development using neuroimaging, much less the much harder problem of classifying and decoding the contents of our thoughts. Slightly more impressive results have been achieved in studies collecting data from actual brain implants found in epilepsy patients, but even then, I do not see this as something that can be a reliable technology in a long long time.
- SmokeGS 7y ago
  I appreciate someone in the field taking the time to respond. I think we have the ability to scale our knowledge with machine learning and mass monitoring. Again i recognize that 50 years is extremely optimistic. Ethics, money, and politics never agree.
  - feanaro 7y ago
    The poster before you is presumably talking about physical limitations of the current measurement methods which leads to very high noise. I doubt machine learning can overcome that.
    - SubiculumCode 7y ago
      I think there are several issues. 1. Our understanding of the brain is poor, and I think that will be necessary to have good machine decoding. In principal, a machine learning algorithm does not need to know how the brain works to make an accurate decoding prediction, but I suspect understanding will be needed to help construct and constrain machine learning models learn to deal with the myriad of mental states, intentions, emotions, etc that can occur. While we can train classifiers better than chance to identify whether one or another item is observed, remembered, etc, these training are quite constrained to a type of information or class. I imagine those classifiers are hopeless when you start performing another activity or mental state, and the machine would need to recognize the new state and apply different decoders. Second, yes there is too much noise in most available methods (EG, fMRI, MEG) and the information is not dense enough (spatial, temporal resolution) and not specific enough to the type of activity (excitation, inhibition, chemical diffusion, etc). Electrodes have done cool things, but I suspect that even millions of wires would not provide enough information and types of information to provide a full fledged brain interface. However, I'll admit, tech is moving so fast...

hurrdurr2 7y ago

"Treatment is worse than the disease". This is also true in cancer treatment, where many drugs have measured effects on reducing tumor size and yet not increasing survival times...yet the side effects make the patients' quality of life significantly worse.

polyfractal 7y ago
I don't think this is fair for several reasons. First, cancer treatment drugs are tested heavily in-vitro first, where the only significant metric is reduction of tumor mass. Clinical trials for cancer treatments are different from other drugs, as you don't generally do double-blind trials (would you want to get a placebo if you had a terminal cancer? No, probably not). So many cancer treatments go into clinical trials with only the evidence that it shrinks tumors, and tends to kill tumors faster than it kills the rest of your body. Side effects are expected to be severe since most cancer treatments are quite literally killing any cells that divide... tumors just divide faster and thus get hit harder. The end result is cancer treatment often has very poor quality of life. Clinical researchers _do_ perform analysis to see if survival increases relative to other treatments. But often you don't know until you've already tried it on a cohort of patients (with their consent, obviously). There is push-back in the medical field in areas for what you suggest. For example, colonoscopies are becoming widely recognized as causing more problems than they solve. The complications they can induce tend to lead to a poorer patient outcome than missing colon cancer because you didn't screen (e.g. survivability is not affected by finding/treating colon cancer early). I just don't think it's fair to paint cancer treatments with the same brush strokes, since dealing with terminal diseases is a totally different ball game. Edit: I should say, there are some parts of cancer treatment which is moving towards the dont-treat opton. For example, there is growing evidence that treating prostate cancer is not necessary, since you'll probably die of something else long before prostate cancer kills you (it is very slow growing on average)
- cannonedhamster 7y ago
  The other problem with cancer is that there are different types of cancer cells. There's the fast dividing and there's also the cancer stem cells which can lay dormant for years without dividing. This is the reason for most cancer returning. There's a promising new tech that hopefully pans out that activates these cells so standard chemo can attack them. Regarding clinical trials, they are very selective. You have to for a very specific criteria and your cancer needs to be measurable by a CT scanner typically to prove efficacy. Add to that the typical cancer patient is either a child or an older person and you're bound to have poorer results.

epmaybe 7y ago

Good luck figuring out how to sustainably pay for personalized precision medicine across the board.

nitrogen 7y ago
Simple: keep doing more of it until it's cheap.

ppod 7y ago

>This testing exists today, but it’s only used in patients who show strong resistance to pharmacological intervention. It should be standard. Is it possible to do this kind of thing oneself with 23andme style SNP tests and a google scholar search, or does it involve more than that?

derekja 7y ago
only slightly beyond that, 23andme doesn't cover all of the relevant SNPs but it does cover the ones that allow you to infer that you are less likely to respond to p-glycoprotein based drugs like citalopram and paroxetine. You have to export your 23andme raw data and use something like promethease to reprocess the data.
JamesBarney 7y ago
No, the poster is seriously over estimating the effectiveness of genetics testing.

DuskStar 7y ago

Isn't this mostly from companies like GeneSight? Considering that they relied on 5-HTTLPR and HTR2A, which both had enormous edifices of completely useless (and now debunked) scholarship behind them, I wouldn't bet on those tests' effectiveness. (For reference, GeneSight's test relied on 7 genes. The five not mentioned deal with liver enzymes) https://slatestarcodex.com/2019/05/07/5-httlpr-a-pointed-review/ https://slatestarcodex.com/2019/05/07/5-httlpr-a-pointed-rev...