Is the Future of Medicine Already Here?

[Chris Riccard]

Anywho, I'm posting today to let you guys know what I'm thinking about. Mainly the future of medical science.

I got the idea by reading a new kind of science by Wolfram, some genetics stuff, some stem cell stuff, developmental bio and chaos and fractals as well as using what I already knew about computers from my electronics background with some computer science added.

So here goes, this is the future, imagine a computer that models (or you could say grows) a baby from a blastocyst inside a controlled womb, (later on you can just use a mother and father genome but let's not jump the gun). The human genome is mapped already but what we have is a very hi level view of what each codon does, when you think of each Gene does it seems like a trial and error way of figuring out the mapping to spacific endpoints with splicing.

Now I'm figuring that if I can I'll probably devote my life to figuring out the structural algorithm of dna scaled (on all levels) proteins to the whole of the anatomy (the scaffolding from dna to complete human in algos (my mathematical intoision is screaming at me with this thing I came across (I hope) called fractal expansion, thinking of protein structure) .

What this will mean is that a computer will be able to have some cheating (mathematical shortcuts) enabling it to make drugs for things like cancer based on the genetic fingerprint of the patient, this will cure most kinds of cancer in the future as well as cheat at Neuroscience because we'll have a complete and mapped out human brain. (Imagine that!)

And that's just the beginning! (Had to add this: with a teachable brain all that's required is I/o , think eyes, ears, nose, voice and sense perception and now reality of human ai is here, the human brain interfaced with a computer!) One more possibility.

 

[jedishrfu]

What is your question? Everything you've mentioned has been talked about in science fiction and medical futurism articles.

Are you thinking of majoring is medicine? and want to know more about these topics?

The notion of computers finding a shortcut is basically the P = NP problem in the context of medical research. Protein design and folding are complex tasks and while we've made great progress in them, we have yet to fold an amino string to near 100% accuracy.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2443096/

While we may be able to construct a brain at some point in the future. I'm sure it take a far longer time before we figure out how to transfer or replicate someones brain into a new brain. Neural connections and neural memories don't come out the same for each person.

 

[Ygggdrasil]

The human genome is mapped already but what we have is a very hi level view of what each codon does, when you think of each Gene does it seems like a trial and error way of figuring out the mapping to spacific endpoints with splicing.

Codons are part of protein-coding genes, which make up only about 1% of the human genome. Somewhere between 10-20% of the human genome is evolutionarily conserved. Therefore, understanding the protein-coding genes is not sufficient for a complete understanding of human genetics; there is at least an order of magnitude more non-coding DNA that plays regulatory roles in determining when and where each gene is expressed. These non-coding regulatory sequences also define the many genetic regulatory circuits inside cells that enable the cells to change their behavior in response to different stimuli. Research into understanding how these non-coding DNA sequences work is still very much in its infancy compared to our understanding of protein-coding regions (which is still very much incomplete as well).

as well as cheat at Neuroscience because we'll have a complete and mapped out human brain. (Imagine that!)

Scientists have known the complete neuronal wiring diagram of the small roundworm nematode C. elegans since the 1970s. Has this given us a complete understanding of the neurobiology of this organism?

some critics point out that the C. elegans connectome has not provided many insights into the worm's behavior. In a debate* with Seung at Columbia University earlier this year, Anthony Movshon of New York University said, "I think it's fair to say…that our understanding of the worm has not been materially enhanced by having that connectome available to us. We don't have a comprehensive model of how the worm's nervous system actually produces the behaviors. What we have is a sort of a bed on which we can build experiments—and many people have built many elegant experiments on that bed. But that connectome by itself has not explained anything."

https://www.scientificamerican.com/article/c-elegans-connectome/

Understanding the human brain will require much more than simply mapping out all of the neurons (which itself is a massively difficult problem).

 

[Drakkith]

What this will mean is that a computer will be able to have some cheating (mathematical shortcuts) enabling it to make drugs for things like cancer based on the genetic fingerprint of the patient, this will cure most kinds of cancer in the future

Cancer cures will probably depend less on your genetic makeup and more on figuring out how to selectively target cancer cells for destruction while leaving healthy cells intact and unharmed.

 

[Chris Riccard]

Let the computer do the work
Mapping out the brain, that's what it's good for.

As far as I know there are 4 ways to attack cancer, one of them just happens to be pre programmed cell death via a RNA vector attack using designer drugs.

The others are, radiation, chemo and the immune system.

 

[Ygggdrasil]

There are many cautionary tales in applying AI and machine learning to problems in medicine. For example, consider the case of the MD Anderson Cancer Center in Texas and it's collaboration with IBM's Watson supercomputer (of Jeopardy fame):

It was one of those amazing “we’re living in the future” moments. In an October 2013 press release, IBM declared that MD Anderson, the cancer center that is part of the University of Texas, “is using the IBM Watson cognitive computing system for its mission to eradicate cancer.”

Well, now that future is past. The partnership between IBM and one of the world’s top cancer research institutions is falling apart. The project is on hold, MD Anderson confirms, and has been since late last year. MD Anderson is actively requesting bids from other contractors who might replace IBM in future efforts. And https://www.utsystem.edu/sites/default/files/documents/UT%20System%20Administration%20Special%20Review%20of%20Procurement%20Procedures%20Related%20to%20UTMDACC%20Oncology%20Expert%20Advisor%20Project/ut-system-administration-special-review-procurement-procedures-related-utmdacc-oncology-expert-advis.pdf from auditors at the University of Texas says the project cost MD Anderson more than $62 million and yet did not meet its goals.

https://www.technologyreview.com/s/607965/a-reality-check-for-ibms-ai-ambitions/

MIT's Technology Review magazine has a nice piece discussing why IBM's Watson platform failed to deliver:

To understand what’s slowing the progress, you have to understand how machine-learning systems like Watson are trained. Watson “learns” by continually rejiggering its internal processing routines in order to produce the highest possible percentage of correct answers on some set of problems, such as which radiological images reveal cancer. The correct answers have to be already known, so that the system can be told when it gets something right and when it gets something wrong. The more training problems the system can chew through, the better its hit rate gets.

That’s relatively simple when it comes to training the system to identify malignancies in x-rays. But for potentially groundbreaking puzzles that go well beyond what humans already do, like detecting the relationships between gene variations and disease, Watson has a chicken-and-egg problem: how does it train on data that no experts have already sifted through and properly organized? “If you’re teaching a self-driving car, anyone can label a tree or a sign so the system can learn to recognize it,” says Thomas Fuchs, a computational pathologist at Memorial Sloan-Kettering, a cancer center in New York. “But in a specialized domain in medicine, you might need experts trained for decades to properly label the information you feed to the computer.”

https://www.technologyreview.com/s/607965/a-reality-check-for-ibms-ai-ambitions/

There certainly have been successes in applying machine learning to medicine (e.g. computers can diagnose skin cancer just as accurately a trained dermatologist), but these successes seem to be confined to tasks that humans can already do well. For tasks that humans currently do not know how to do, it has proven difficult find the right training datasets on which to apply machine learning. There is certainly a lot of promise, but making progress is not just a matter of writing better code.

 

[Drakkith]

You can't map out the brain of a person based on their DNA. The growth of neurons and their connections is somewhat random, with the exact number of neurons and the configuration of their connections not fully predetermined prior to development.

 

[Chris Riccard]

The thing your progming the computer to do intelligent cell division with a known set of rules based in math and relying on dna to show the way, my idea is to start with a computer simulation of an embryo then let the computer (once the structural algorithm is known) build using dna information so this is not like Watson, I'm giving the computer no choice but to follow the rules of dna instruction.

 

[Chris Riccard]

In the far future we'll be able to tag the optic nerve in the computer like a memory address so in theory brain development is possible, imagine a camera or even doing it in the computer directly via virtual nerve data of watching a cartoon.

 

[Drakkith]

Like I said, certain aspects of neuronal development are random, so you literally cannot fully map out a person's brain prior to development. Running your simulation could probably give a close analogue of the person's brain, but not an exact replica.

 

[Chris Riccard]

I wasn't really thinking of making a replica but a unique identity.

 

[Drakkith]

I see. Well, that's certainly a long way off from now, so we'll have to wait and see if it becomes a possibility.

 

[Ygggdrasil]

The thing your progming the computer to do intelligent cell division with a known set of rules based in math and relying on dna to show the way, my idea is to start with a computer simulation of an embryo then let the computer (once the structural algorithm is known) build using dna information so this is not like Watson, I'm giving the computer no choice but to follow the rules of dna instruction.

I see. One important issue here to consider is that genetics do not determine traits 100%. For example, identical twins share the same DNA, but of course, they are not identical individuals. Environment plays an important role in determining how genes express themselves, and as Drakkith has noted, many stochastic events can play important roles in development. Still, an ab initio computer model of an organism (or even a cell) would be very useful for research and would definitely enable a lot of medical advances in the future.

Researchers are working toward creating computer models of simple, single-celled organisms like bacteria (for example, see this PF disucssion https://www.physicsforums.com/threads/computer-model-of-a-bacterium.622587/). However, even in these simple organisms, we still don't know the precise function of about 30% of the genes required for life, which is one impediment toward improving computational models of life.

 

[Laroxe]

Genes in isolation are simply puddles of chemicals, the important things to consider is gene expression and the control of gene expression is complicated. Many of the things that make us human is not the effect of single genes, many of which are shared with other organisms, its the way in which networks of genes interact in a particular environment. Its becoming increasingly clear that our understanding of how genes work is limited at best, even the idea of genes having a single function as in the synthesis of a single protein is wrong.

In humans its suggested that 70% of our genes code for at least 4 amino acids, I believe the current record is held by a fruit fly where a single gene has been associated with over 200 different different products. These are things that we are only just starting to address.

The second problem is in the assumption that the brain can be effectively modeled by computers, while we use computers as an analogy when talking about the brain, that's all it is. The brain doesn't function like a computer and once again our understanding of how it does function is limited and investigation has been guided by ideas that are quite clearly wrong, it might be better to think of each individual neuron as an information processing device, they certainly don't operate as a simple switch. I think this puts these ideas well into the future, there is certainly no immediate prospect of anything similar. Basically we haven't much of a clue.

http://www.iecb.u-bordeaux.fr/index...t-un-gene-peut-coder-pour-plusieurs-proteines
https://www.ncbi.nlm.nih.gov/pubmed/11127834