From Computer Tyme Support Wiki

Latest revision as of 21:00, 29 December 2017

Introduction

We all know what the function of the immune system is. The immune system is there to protect the stuff that we are made out of and to get rid of the stuff that will cause us harm. But how does it work? How does it know who is the good guys and who is the bad guys? How does it tell friend from foe?

There are thousands of papers written about the immune system. They talk about the different cells involved. They talk about proteins, peptides, antigens, antibodies, vaccines, bone marrow, self vs. non self, signaling pathways, apoptosis, and hundreds of other concepts that will make your head spin. This paper is different. This paper talks about the immune system in the context of function and process rather than focusing on the mechanisms of how it works. In this article I compare the immune system to an email spam filtering system which functions as an engineered immune system for email.

The goal here is to look at the big picture and work back to the details. An email spam filtering system and the human immune system ultimately have to accomplish the same things, to identify the good and the bad and to protect the good and eliminate the bad. Although one is a biological mechanism created by evolution and the other is entirely software, on a functional basis they are extremely similar, even though the mechanisms are totally different. Both systems have innate and adaptive mechanisms, where the innate system is fast and rules-based, and the adaptive system is programmable and can learn friend-and-foe dynamically.

The Goal - A new understanding of the Immune System as a Programmable Information System

Most doctors were taught that the immune system protect "self" and rejects "foreign" or non-self, self being what you are born with and non-self is everything else. A very simplistic model. The a researcher Polly Matzinger came along with what is called "The Danger Model" (DM), which says that it isn't about what is self, but what is and isn't causing trouble (danger). It's more accurate, in that it explains why women don't reject their breasts at puberty, or their fetuses during pregnancy. It also explains immune responses to organ transplants, autoimmune disorders, and other inconvenient truths that "The Self-v-Foreign Model" (SFM) couldn't explain.

Certainly DM is intuitive, in that it mirrors evolution itself. That which doesn't recognize danger doesn't survive, and that which recognizes danger does. Dr. Matzinger's new model not only explains what we observe, but it makes predictions that lead us in the right direction to achieving future medical breakthroughs.

This is a new model. It's a model that builds on the work of everyone else, but it has a different perspective. I, Marc Perkel, am not a doctor. I come from the IT world; and I come from the self-educated world. That makes my view different than those in the medical and formally-educated world. And sometimes a different view can lead to insights that one can not have from inside a culture.

What I'm presenting here is very intuitive to people in the IT community, and not as intuitive to the medical community. I remember being in school and learning that whales were not fish. It was hard to accept. As our understanding evolves we are learning that we are more like computers than we thought, and that many principles of digital systems apply to biological systems. Our DNA, for example, is digital, and can be sequenced into a number. That number can, in theory, be transmitted to another solar system where it can be reconstructed in a DNA sequencer, and an identical twin can be made on a different planet.

What I'm trying to do here is to get you - the reader - to understand the immune system from an information technology perspective, and that perspective will lead all of us to new discoveries. This new model makes interesting predictions that can be used to create new therapies and new cures for common ailments. These include:

• In-situ cancer vaccines to cure metastasized cancers.
• An anti-vaccine to cure auto-immune disorders by reprogramming the immune system to learn new friends.
• Inventing new mechanisms to improve the learning algorithms, by adding digital learning to biological learning.

I ask the reader to think differently and expansively, and not let previous learned models of reality interfere with the understanding of a new way of thinking. Allow yourself to go outside your comfort zone, and understand medicine through the eyes of an information technologist, and see the software within the human hardware. See the functions to more fully understand the mechanisms.

The Immune System is an Information Processing System

The main point I hope to get across here is that the immune system is an information processing system. We think of the brain and nervous system as being the center of learning and decision-making. But as it turns out, it isn't the only intelligent system in the body capable of learning. Although the immune system isn't the center of consciousness or self-awareness or thinking, it is a programmable computing device that is capable of learning and making sophisticated decisions.

The Brain is Part of the Immune System

In fact, one could argue that the brain is an integral part of the immune system. When we become sick enough, our brain decides that we should call the doctor, who prescribes and antibiotic that kills the infection that is attacking us. And it's not just limited to our own brain. The antibiotic exists because of the actions of many brains working as a team within a global society to invent, manufacture, and distribute this medicine to those who need it.

Between antibiotics and the invention of the flush toilet, our extended immune system has doubled our human lifespan. Our excretory system used to end at the anus, but today that is extended to the toilet, the sewage system, and the sewage treatment plant. And although these systems aren't physically or biologically connected to us, it would be a mistake to treat these systems and separate from us, just because they don't contain human DNA. My body is a colony of organisms and 90% of those critters are non-human, yet I could not live without them. So are they self or foreign? Does self and foreign even really have any useful meaning? Is a bee a single organism, or is the organism the hive which is made up of bees?

A Vaccine is a Database Update

As part of society and as a function of our brains working individually and collectively, humanity has developed vaccines to protect us from getting infectious diseases. But what is a vaccine?

• Is a vaccine a medicine that kills a pathogen? No!
• Does a vaccine weaken or harm it's intended malicious target in any way? No!
• Does a vaccine affect organs and tissues, or make biological changes to improve immunity in general? No!

When you go to get your flu shot or your tetanus shot, what are you getting? What they are injecting you with is data. You are getting information. A vaccine is a software update, it's code; and it is used only to program your adaptive immune system.

Functional Equivalency Between Biological and Computational Systems

People used to get polio, and now we don't. Why is that? Did humans evolve to be immune to polio? The answer to that question is both yes and no, depending on your definition of what constitutes human evolution. If you understand human evolution narrowly in the context of our biology, then the answer is no. If you take an unvaccinated person and expose them to polio, then they get polio. However, if you understand human evolution expansively, then who we are doesn't end at the surface of our skin. It includes our technology. Using our technological immune system, we vaccinate people to be immune from polio, and therefore we no longer get polio. So humans have evolved immunity from polio.

I contend that the expansive model properly describes human evolution. It is certainly more compatible with the goal of improving life through the use of technology. Is your computer self or foreign? I contend that your computer is self.

Many of us run antivirus software on our computers. The software has two components. It contains code that identifies and eliminates foreign software that is causing trouble. Some of this might be identified automatically based on the behavior of the software. But mostly it works off of a database that contains identifying signatures that it uses to match known troublemakers.

When you get a newly written virus, your computer might not recognize it at first. But then you get a database update that contains the signature of the new virus, and now it sees the virus software as enemy, identifies it, and eliminates it. Similarly, if you are not vaccinated and you're exposed to polio, the polio attacks you. But if you get the polio vaccine you are getting chopped up pieces of polio that your adaptive immune system can recognize. These polio pieces are injected with an adjuvant, a chemical that classifies the polio as enemy, and your immune system updates its database and learns polio, classifying it as enemy. Then when real polio comes along, the immune system responds and stops the polio before it can get started making trouble.

Information Processing Model Leads to a Deeper Understanding of Immunological Theory

Although the mechanisms of computer antivirus software and the immune system are totally different, they are functionally equivalent. The computer identifies digital patterns (recognition) with a classification as enemy. Similarly the adaptive immune system identifies protein sequences (recognition) and the adjuvant classifies these proteins as enemy. They do the same thing.

Because we are in the computer age, we are required to create an artificial immune system for our computers to protect them from malicious invaders that would destroy functionality. This gives us a deeper understanding of the processes necessary to accomplish this task. Because of this understanding, we can apply these insights to the human immune system to more fully understand similar functions that have to be accomplished in order for a similar process to work. And this deeper understanding creates a big picture that ties the mechanisms together so that they whole thing makes sense in a greater context.

Comparing The Immune System to a Spam Filter

Th immune system and email spam filters actually do the same thing. The job of a spam filter is to look at a stream of inbound emails coming from the internet, examine them, and classify them. If the email is good it is forwarded on to the recipient. If the email is bad, it is blocked. Similarly, the immune system examines the body's own cells - friendly organisms that live with us, invaders, pathogens, bacteria, viruses - and decides which things are friends, and which things are foes. The friends are protected, and the foes are eliminated.

Here is a functional diagram of a spam filter.

                 +---------------+
                 |               |-----[good email]------> delivered
-----[email]---->| Spam Filter   |
                 |               |-----[bad email]-------X blocked
                 +---------------+

And here is a functional diagram of the immune system.

                 +---------------+
                 |               |-----[good cells]------> protected
-----[cells]---->| Immune System |
                 |               |-----[bad cells]-------X elimiminated
                 +---------------+

As you can see the overall function is very similar. But let's break it down further, and look at the details of how it happens. The immune system is loosely divided into 2 parts: the innate immune system, which is hard-wired to recognize the easy stuff; and the adaptive immune system, which is programmable and can learn to recognize the tricky stuff.

Similarly, the spam filter has a rules-based sorter that can identify email as good or bad based on a number of easy tests - the source of the email, behavior of the sender, content of the message - and it quickly classifies most email as good or bad. But it also flags a lot of email as "unknown." which requires a smarter system to determine classification. The spam filter feeds these unknown emails into an artificial intelligence engine (AI). It makes the final call to classify the tricky ones.

                 +----------------+
                 |                |-----[good results]-------------+------> protected
                 |                |                                |
                 |                |     +--------------------+     |
                 |                |     |                    |-->--+
-----[input]---->| Easy Selection |-->--| Adaptive Selection |
                 |                |     |                    |-->--+
                 |                |     +--------------------+     |
                 |                |                                |
                 |                |-----[bad results]--------------+------X eliminated
                 +----------------+

Both system are similar in that they have a dumb-but-fast classifier that does most of the work, and a smart-and-programmable system to handle the tricky cases. Although in real life both system are not as distinct as depicted here, this model fundamentally conceptualizes how these systems work.

How The Adaptive System Learns

In both the spam filter and the immune system, the adaptive system has an artificial intelligence learner. In the spam filter, that learner is a digital database that contains pieces of email used to match against incoming email. In the immune system, we have a chemical database that consists of stored protein sequences called antibodies, which can be used to identify antigens on the cells surface. Although the mechanism of these two systems is very different, the functionality is the same.

Although the information is different in each kind of database, each database entry contains two parts: an identifier (which is used to match the unknown sample), and a classifier (which is used to label the match as good or bad). The samples, whether email phrases or protein sequences, are matched against the good and bad databases, and whichever side has the most matches wins the classification.

                 +----------------+
                 |                |-----[good results]-----------+------+------> protected
                 |                |                              |      |
                 |                |                              v      |
                 |                |          +--------------+    |      |
                 |                |        +-| Good Samples |----+      |
                 |                |        | +--------------+           |
                 |                |        v                            |
                 |                |        |                            |
                 |                |     +----------------------+        |
                 |                |     |                      |--->----+
-----[input]---->| Easy Selection |-->--|  Adaptive Selection  |
                 |                |     |                      |--->----+
                 |                |     +----------------------+        |
                 |                |        |                            |
                 |                |        ^                            |
                 |                |        | +--------------+           |
                 |                |        +-| Bad  Samples |----+      |
                 |                |          +--------------+    |      |
                 |                |                              ^      |
                 |                |                              |      |
                 |                |-----[bad results]------------+------+------X eliminated
                 +----------------+

The innate/rule-based filtering provides the information which trains and programs the adaptive/AI systems, providing both with information, and teaching them, "This is what good and bad looks like". One can look at this training as a moral compass. It allows the adaptive systems to get a sense of right and wrong. Thus, when presented with an unclassified sample, the adaptive system determines if it looks more like the stuff classified as good or bad.