This story is over 5 years old
Tech by VICE

Let's Apply Network Analysis to Medicine

This is how I dream of the future: People can go get a brain scan—maybe we have disposable imaging headpieces at this point—and then connectome translators interpret their brain scan into a list of personalized features.

by Kara Crabb
Sep 19 2013, 2:10pm
Photo via Flickr / CC. 

The translation process that makes scientific research tangible to the medical field takes a long-ass time. In many industries, actually, the social engineering required to implement things that are waaayyyy more effective to them is just not there. In computational neuroscience there’s evidence to say that people can benefit from, at the very least, biomarkers for disease that emerge out of network studies. This is just one reason why the “connectome” might actually be useful.

In any network—be it the internet or the human brain—disruptions that occur in major “hubs” cause the entire system to suffer. For example, if one of your shitty Facebook friends deletes their account, nothing really happens. But if LaBron James decides to boycott Twitter, a lot of activity would be lulled. That’s basically the difference between a highly connected neuron, or cluster of neurons (?) versus a neuron that doesn’t see much synaptic traffic/ that’s the difference between a “hub” and a “node”.

Obliviously in neurological diseases and mental disorders, different patterns of vulnerability will occur within the systems that are effected by them. As more information about the details of hub disruptions become available, it’s progressively clearer to conceptualize neurological diseases like autism, Alzheimer’s, and schizophrenia, as network challenges since very specific, targeted irregularities are taking place.

The metabolic and molecular processes that are involved in targeted irregularities correspond with the global network architecture (ie, the brain.) For example, the efficiency of a circuit, which helps makes up a module, could suffer from factors like myelination or axon length or protein deposition. So from micro to macro scales, information gained from network analysis all fits together. In this regard, maybe a good goal would be to design drug-like molecules that execute localized effects. It’s possible!

One of my favorite supporters of computational neuroscience is Norbert Weiner. He was a famous child prodigy and mathematician. In the 40s he wrote about cybernetics and psychopathy. In the 40s. He grew up with a younger brother who had schizophrenia, and was also acquainted with game theorist John Nash. He thought that the flow of information through the nervous system was vulnerable to overloaded pathways that could ultimately lead to mental breakdown. It turns out that this idea proposed is actually just gaining momentum today. Norbert Weiner, you are my spirit animal.

Schizophrenia encapsulates so many different cognitive and behavioral symptoms, there’s no way I can give it justice with a succinct definition here. Also, this is an important note: An objective diagnostic test doesn’t even exist yet. One large issue right now is that there’s no noninvasive way to record neural activity at high spatial and temporal resolution. Most methods measure activity indirectly, and each method has its own advantages and disadvantages.

via Flickr / CC. 

As it stands, it seems that in order to get a clear picture of one schizophrenia diagnosis, the patient would have to go through several different imaging techniques, which would be exhaustive and expensive and probably not do very much for them anyway since the disease is so obscure. The most basic conclusion that has come out of schizophrenia research is that abnormal connections are taking place. There are also structural disturbances in white matter, apparently.

To me—and I have no education so this doesn’t matter—network analysis really sounds like an integral diagnostic tool for schizophrenia, in particular, because it sounds like variations of the “disease” occur ephemerally and frequently in people who don’t actually have it.

Alzheimer’s has been straight up described as “cortical disruption syndrome”. Although the molecular mechanisms of the disease are still unknown, researchers have been able to track patterns of amyloid and tau deposition in AD patients, which leads to deterioration. Highly central regions are the most vulnerable to atrophy, which I’ve always found kind of funny. 

In particular, one of the first regions to break down is the medial temporal lobe, which is a region associated with memory (kind of vague, but I’ll take it.) Recently, amyloid depositions have also been found in a lot of normal older adults too. So, it seems like right now there’s a lot of crossover between early AD symptoms, dementia symptoms, and just straight up bein’ old. I think network science is special because it could help organize these features into proper clinical profiles.

Neurodevelopmental disorders like autism and ADHD also have documented, regionally-specific attributes. Children with autism have an increase in both grey and white matter, white matter pathways apparently being “disorganized”. They also have an overabundance of short-range connections paired with a loss of long-range global integration, which could explain why kids with autism are “really good” at “certain things”. Personally, I hope we can figure out how to actualize autism and make it a reality for people who haven’t been blessed with it from birth. Is that so wrong? I would totally trade my charisma for prodigy-status, Jacob Barnett steez.

It seems like a no-brainer that network analysis belongs in medicine immediately. This is how I dream of the future: People can go get a brain scan—maybe we have disposable imaging headpieces at this point—and then connectome translators will interpret their brain scan into a list of personalized features. Hub disturbances or risks would be listed, and people could look at the list and then order designer drugs to alter the shape of their malfunctions, or even, dare I say, to slightly enhance connections (for a certain amount of time?) if they wanted to.

Society would be the same, because people would just chose how dumb or cool or crazy they wanted to be. Neuropathology wouldn’t even really be a thing anymore. Oeople would just be like, “Oh, one time I had autism but now I’m talented AND charming.” Then we can all blast off to space before the planet kills us, or whatever. Isn’t that why people do things? Or something? Good talk.

network analysis