Divergents and Directional Antennae: Measurement and Inference

λ  (  §µßƍŁ¥þĦ ԠƎ  )

★ Part One: Divergents and Directional Antennae

In Which David Conner Cures Cancer

This article continues from the last one, but is concerned moreso with analytically modeling a system from observed measurements. Then I continue on to establish the concept of neurological & neuropharmocological “fingerprints” by which diseases can be diagnosed earlier via MRI. And by looking for patterns and deviation from patterns, we should be able to treat cancer much, much earlier. *(This section header is facetious; many of these techniques are already known, but the ramifications to available cancer diagnosis methodologies are serious)

How Many Experimental Measurements?

Continuing from part one, how many experimental measurements does one need to make in order to infer the distribution of EMR energy from an antenna? Is it less or more? For a quadratic equation, you need 3 measurements. For a cubic equation, you need four. And generally, for any n-polynomial in two dimensions, you need n+1 measurements.

But for determining this kind of distribution, how many measurements do you need to make to be more confident in the rest of the values? When determining the average value for EMR in any SVE over a short time-scale? And when you take into account how the values for each SVE and AVE fluctuate over time? Note: SVE is spatial volume element and AVE is antennae volume element from the model in part one)

Is there the same kind of relationship as with n-polynomial equations in two dimensions? Where more information allows you to confidently make more inferences? Or is there some maximum confidence with which you can infer the distribution because of the nature of the mathematic model? Can you fluctuate the power of the antenna when you make measurements and use the derivative of EMR value with respect to the derivative of the power fluctuation to make better inferences? I’m pretty sure you can.

Noise Confers Intuition

On the surface, it would appear that you have to make more measurements the more complicated the relationship is, but I contest that the more complicated the relationship, the more specific it’s influence – and therefore, the easier it is sometimes to make more confident measurements.

And especially, if you can inject some known pattern of noise into the system, you can generally make inferences that allow you to more confidently understand the nature of the system, even though “it’s just random noise.” This is especially true if the derivative is useful because you can look for the signal of the derivative of the noise transformed in the output.

When Methodology Affects Data

When the methodology affects the data you collect, this makes the inference of something like the distribution of EMR so much more difficult. This is because you can never truly measure the thing you want because the instrument or whatever introduces noise that changes the values. So you’re measuring a system that’s completely different from the one you want to understand. It basically turns your systems into something like differential equations.

However, the analogy to linear polynomial equations in two dimensions above applies here: you’re just comparing systems of equations, not values for points. So, to understand the “zero system” which is like the system of equations which you haven’t actually affected with your methodology for measurement, you just have to observe enough kinds of “non-zero systems.” Then you should be able to cancel things out and infer the nature of the “zero system” you’re looking for.

Statistical Resolution

When you add the “dimension” of statistics to your mathematic models, you are empowering yourself to understanding the likelihood of a system to assuming various states. This allows you to solve the problems from a completely different perspective. You almost don’t need to know anything about how the system works to begin reducing it further. You’re basically just eliminating space until you find an analytic system that at least simulates the numeric system you’re looking for.

This is similar to how the LHC engineers looked for the Higgs Boson. They kept making measurements with experiments at high enough energies until they reached sixth sigma. They almost certainly eliminated the possibility of a particle with energies of the Higgs not existing.

Artificial Intelligence

But you can pretty much apply this statistical dimension to any system and whittle away at the space of things that are not true, leaving you with a better understanding of what must be true. This scares the shit out of me, in the context of AI. It basically means that the more information and data that AI has to process, the more completely it can understand a system and the more efficiently it can do so.

This can also delude people, who, when given a deluge of information about a system, can be led to making the wrong decisions about it by hyperfocusing on one subsystem within it. E.G. when people have security cameras, they feel safer because they think they see everything going on, but often their eyes lie to them. But this doesn’t really work with AI because machine learning algorithms can process information, the significance of which it doesn’t need conscious awareness. Yes, bias is a huge problem in machine learning, but given enough processing power and enough of the right kind of data, it will always produce frighteningly accurate inferences about systems.

Noise Injection X Statistical Imaging

This noise injection, as I have noted on twitter, is very useful in understanding the results of MRI, especially diffusion-tensor imaging. We should be able to conbine data from various methodologies.

The following inferences do not cite sources. Do not infer what I write here as fact. Do your own research. Keep reading if you’d like to know how to cure cancer. And by cure, I mean identification through such early and non-invasive methodologies that it rarely reaches the later stages like metastasis, which are notoriously hard to treat.

Spatial Coherence

First, to link images spatially; to get spatial coherence between image sets. There are several types of MRI – T1 and T2 is one distinction. Whether or not a dye was used is another dimension. There are sagital, axial, and coronal image sets. It’s all lined up with the same physical space as a domain generating the MRI data. Though the individual spaces of each MRI methodology may be different, they can all be correlated to the same 3D cartesion physical space, relative to the person being imaged. This means that specific structures in the brain, varying in how the data specifically depicts them, can be correllated across images.

This technique can link multiple types of MRI images, so they can be corrected into the same spatial coordiante system and their values can be combined. Their values would be spatially shifted though, so standard algorithms like binomial image interpolation would not work without some customization. However, the higher resolution the MRI, the less effect that spatial dissonance would have on image analysis, since data from the images would be more spatially “in phase.”

Temporal Coherence

Temporal coherence varies from short term to long term, but it basically means the identification of patterns in how systems evolve over time. Short term: being able to track blood flow through someone’s brain or change in oxygenation of specific structures in the body, with respect to specific body postures – ahem: yoga. Long term: understanding patterns of development in brain structures or differentiated brain development of people who lack sensory input. By the way, if there really is a sixth sense, and someone seems to have it very strongly, then analyzing neural structures for anomolous development progression types may allow us to identify a physical structure that enables E.S.P. Ahahahh, that’s probably not real though.

Categorical Coherence

One final step is to get categorical coherence, which requires understanding more about the input in order to make inferences across data sets. For MRI, this means understanding the patient’s age, recent/chronic diet/nutrition habits, medical status, etc. So that inferences can be made across data sets. I.E. without spatial or temporal coherence, the hippocampus could be recognized in an axial image by an algorithm trained to identify hippocampus shapes at various ages, young or old.

This “categorical coherence” looks for higher level patterns. It could identify the presence of diseases previously unknown, not because it’s matching against a list of diseases and their progressions, but because it’s identified unknown patterns of neurodegeneration in a patient. In addition to unknown disease progression, this could also identify the combination of drug-induced neurological damage or damage induced by the combination of diseases. It could even identify WHAT DRUG CAUSED THE NEUROLOGICAL DAMAGE by understanding the dispersion and extent of damage in the brain.

Neural Fingerprint for Pharmaceuticals

Drug Absorption in the Human Brain.

The brain is very particular in that it has the blood brain barrier, which restricts entry of charged particles. The blood brain barrier is a linining of brain cells that receives and filters nutrients from arteries and capallaries. The only way a charged particle can pass through that is if it is briefly neutralized, which is a low-probability event. The point is that this has a big effect on the dispersion of specific chemicals throughout the brain, which form the basis of a “neural fingerprint” for neurological damage or for normal activity and absorption.

Neural Fingerprint for Receptors

Another factor that influences dispersion of chemicals throughout the brain is expression and distribution of the receptors a drug interacts with. Because our brains evolved to interact/reuse chemical messangers that share similar qualities, structures and behaviors, our genes developed and reused long sequences of amino acids and began to copy/paste those sequences elsewhere within the genome. There are many introns within genes which are reused or similar versions are pasted elsewhere.

This allows chemical behavior to be encoded in one place, but reused. It exponentiates the variety available to our genetic programs. It’s analogous to an input/output adapter, like a 1/4” to 1/8” headphone adapter. It’s also analogous to interfaces and inheritance in object oriented programming. The point is: our bodies evolved similar mechanisms to work with similar chemicals. And particularly, to work with different spatial aspects of a specific chemical messenger’s chemistry. This is why some SSRI’s have such a wide array of recepter activity. Different parts of the drug fit different locks. And this is true with engodenous chemicals too.

Blood Flow and Neuropharmacology

The point is, when someone takes a drug, it flows in through the arteries and hits the places that receive the most blood first. It then reaches the capillaries. But if it’s a charged particle, it takes forever for it to penetrate the blood brain barrier, if it’s at all possible. This is because the molecule needs to undergo several chemical changes, which are low-probability events. And it can’t even get into a neuron (or affect the surface ion channels & receptors of a neuron) until it does. This is why strong acids and bases are incredibly dangerous – JESUS CHRIST CUTTING COCAINE WITH BAKING SODA IS EVIL. Strong acids and bases could damage the blood brain barrier and cause hellacious damage. I don’t know enough of the structure of the endothelial cells in the blood brain barrier to know if it’s a significant problem, but it seems like that could be damaged by strong, localized deviations in pH.

By the way, cigarette manufacturers include ammonia because it changes nicotine into its “crack” form. The freebase form of nicotine bypasses the blood brain barrier faster and is felt almost immediately.

The receptor profile determines where a drug will stick around. The receptor profile is a list of receptors, along with information on whether or not a drug will activate it, inhibit the action or block other chemicals from interacting with it. It also includes the timescales for drug interaction with receptors. Some drugs bind tightly to receptors for much longer than others, blocking their action. The amount of time a drug interacts with a receptor, along with the receptor’s prevelance in areas of the human brain determines how long a chemical will remain in an area of the brain once it’s penetrated that region via blood flow. If those receptors have already be activated by another drug that binds more tightly or for longer, then the molecules will continue randomly floating around until they find something to interact with. The pattern of space in the human brain that a drug would normally bind to forms a kind of neuropharmacological fingerprint.

This is incredibly complicated and hard to measure. And leads into another subject on predictive medicene, which I will write on soon. The point here is that the spatial information trumps genetic information. The distribution of receptors matters a lot more than the phenotype of the patient’s genetics. And again, it is much harder to measure!

For some CYP450 enzymes in the liver, their relative quantities could be measured by observing plasma concentrations of a drug metabolized by them, but that’s invasive. And still, there are many drugs which are acted on or interfere with the action of multiple CYP450 enzymes, making it difficult to ensure accurate testing. If your blood cholesterol tests are inaccurate because of recent diet changes, quantifying individual CYP450 levels would be two or three orders of magnitude harder.

Competing Chemicals Change the Neural Fingerprint

One more factor is the presence and absence of competing chemicals and their profile for neuological activity. These competitors can preclude other chemicals from interacting with receptors. If a chemical has a stronger binding action with a specific receptor, it can knock other drugs off and interupt their activating or inhibiting or blocking action. This is regardless of how long they stay bound to the receptor, even though strongly binding drugs tend to stay bound for longer. This is not always the case.

The action and interaction of competing chemicals causes their neural fingerprints to shift. This isn’t very predictable, as it is strongly dependent on the distribution of receptors in the human brain, as well as plasma concentrations of endogenous and foreign chemicals. But that distrbution of cell types typically determines distribution of types of receptors, especially in the brain. So if it’s possible to use machine learning to understand the location of neuronal types in someone’s brain, given simply an MRI, as well has higher order structures like the Substantia Nigra, then this could lead to novel diagnostic criteria. These methodologies could predict cancer and disease much earlier and much more accurately than many of the current methods. And not just in the brain!

Identifing Neuronal Structure Types

So, if you can pinpoint the Substantia Nigra and you know that it often contains Nicontinic Acetylcholine Receptor (nAChR), then you can assume that the presence of these receptors and the specific shape of this region of someone’s brain may affect the distribution of nicotine. Also, if you detect that someone’s substantia nigra or some other structure is significantly distorted, this may be caused by strong local pressure in their brain, which could be an indication of a brain tumor. Just by using algorithms to classify types of structures in the human brain, we can identify them and quantify their deviation from specific shapes.

This data can be combined with Diffusion Tensor Imaging, which is a powerful way of analyzing neural structures. Basically, the DTI measures how water molecules vibrate under strong magnetism. If they vibrate with a mostly random, 3D, spherical dispersion, algorithms can infer there is no structure there. But if the molecules vibrate with a linear dispersion – that is, if their pattern of dispersion tends to be stretched out in a long ellipsoid, then algorithms can deduce there must be some kind of neural fiber there. DTI algorithms can deduce higher-order structures in the brain and help identify abberantly connected regions in connectopathies like autism and schizophrenia.

Ariel Rokem - DiPy Presentation on DTI

Curing Cancer

With Early Prediction and Detection

This cell type and shape distribution may allow us to identify other kinds of cancer earlier. This is much more effective than predictive medicine from genetics, which seems to be what everyone else is working on, because, instead of needing to understand how someone’s genes have been expressed in their lifetime, it rides on top of that. For example, if you synthesize a drug or pharmaceutical with isotopes that react in an MRI and administer that to someone being imaged, then you can watch out the distribution of that drug changes over time and make assumptions about the size/scale/distortion in structures throughout the body. Not just the brain. This means there are lots of diagnostic methods for cancer prediction that we are not using.

Back to the thing I actually posted on twitter – 3000 words later… You can do this with water! It’s actually been done for a few years now. Hydrogen and tritum resonate with Nuclear Magnetic Resonance (NMR) But deuterium doesn’t and therefore disperses differently throughout the brain. Especially as time goes on. That means you can take MRI images with natural levels of deuterium, using those as control. Then titrate deuterium into someone’s brain and watch out it interacts. The stronger the magnetization of the MRI, the better. Most MRI’s are 1.5 Tesla or 3 Tesla. 6 Tesla would better.

As the deuterium penetrates into neuronal structures in the brain, you get a better idea of the shape of the brain because you’ve basically injected noise into the system. The deuterium doesn’t resonate and therefore the water vibrates less in areas with more deuterium. You can observe the path chemicals would take as they penetrate someone’s brain because you may be able to measure the local deuterium quantity as the water begins to vibrate less and less in those areas.

Furthermore, you can use a little known effect called Magnetization Transfer (MT) to make even more inferences with the action of Deuterium. This occurs when hydrogen ions vibrating near protein and macromolecules swap with hydrogen atoms in those structures. And (I think) when this it gives of a specific radiological reaction, which allows it to be detected as an anomoly. Well the behavior of that anomoly should change in the presence of deteurium, which should lower the rates of magnetization transfer.

I don’t know if these anomolies produce signal which is strong enough to be categorially determined as specific protein structures, but some of the math from XRay Crystallography may prove to be useful here. Over larger regions, the anomoly of signals from magnetization transfer should be repeated in some way, across space. That is, the signal from magnetization transfer to specific cell structures should acquire a specific quality. And if those cell structures are repeated for a neuron type that includes specific receptors, then you may be able to map out those cell types or even quantify the amount of those receptors.

…… But probably not. I don’t know how much of a benefit magnetization transfer has to offer. I know that too much deuterium is poisonous – 20% of body weight. And I know that if you use this methodology, then a person will absorb the deuterium, basically introducing long term noise.

Back to Neuronal Fingerprints

But if a drug should have a specific finger print, the by analyzing that fingerprint (even decades later) you should be able to determine whether someone was exposed to a specific neurotoxin. Or if one specific neurotoxin happened to be applied accross a large population because it seemed to have attractive pharmacological qualities as something that could be masked or hidden, then that could be identified without needing a hair test or much proof at all.

Does It Really Matter What I Do?

If I do have a right to be mad at the world for what has happened to me, which I still don’t entirely understand, sometimes I think the best thing I could do to get back at them is to kill myself. I can do so much for people, but I made $1717 last year and I live in America. I have been effectively blacklisted and I seem to be systematically harrassed wherever I go. Why should I want to help a population of people that would do that to me, if that is really what is going on? Is it? Jesus Christ, that’s crazy as shit.

But the point is: it doesn’t seem to matter how smart or talented you are. You can’t just “be” the way you want to be. You have to protect and shield and enshroud and obfuscate your talents as though they are something that must be hidden from others. Especially if they are the kind of thing that could change the world. Because if you do not protect yourself by artifically contorting your situations and environment for profitability, you will find yourself robbed and derided as nothing. As a worthless failure.

It’s really sad. It’s so easy for many people to get filtered out if they don’t conform. But the nature of knowledge is such that for any piece of it to be replicated, it only needs to be incontrovertibly demonstrated once. From that point, it can be restated and replicated as truth, on the basis of faith alone. Yet, after that, the person who reached those conclusions is no longer necessary. Perhaps, even less necessary than before. In some circumstances.

That attitude; I don’t believe that’s right.