Exploring A Network Map of Neurotech I
Right now I am exploring a map of 4,000 articles about neuroscience technology.
This post follows how a complicated network was created over many hours, from keywords to queries to visualization to exploration.
I use Quid to map complex networks of information. It can currently handle 10,000 nodes. Nodes are entities — in this case, articles. Selecting a mere 10,000 articles to characterize a world of neurotech is no small task.
We started with over 920,000news articles published between Aug 2013 and Aug 2014. Together, they encompass 25 keywords (original query quoted below) designed by MIT to capture coverage of neuroscience technology at a global level.
“brain computer interface” OR “brain machine interface” OR “brain imaging” OR EEG OR “cell engineering” OR “cognitive control” OR connectom* OR DTI OR “electron microscopy” OR “fMRI” OR “genome engineering” OR “high density electrophysiology” OR “machine learning” OR “materials science” OR “magnetoencephalography” OR microfluidics OR “advanced microscopy” OR “magnetic resonance imaging” OR plasticity OR stimulation OR neuroprosthetics OR optogenetics OR “structural MRI” OR synthetic OR whole brain optical imaging OR neural circuit modulation AND neuro*
Getting to under 10K is kind of like protein combination. You make a series of 19 smaller networks, each of which singularly characterizes a specific neurotechnologies. Then you merge those together into one master.
Defined AND NOT
Some queries are as simple as
“electron microscopy” AND neuro* NOT fellowship* NOT funding = 1105
Others, such as that for magnetoencephalography, a technique ubiquitously written as ‘MEG’, requires more creative filtering
“MEG” AND neuro* NOT market* NOT “Meg Ryan” NOT “Meg Loncharic” NOT “philharmonic” NOT invest* NOT “Burlington” NOT “Ice Bucket Challenge”
Then there are well-reported technologies, such as MRI or EEG, returning 10K or 50K results.
Filter by News Genre, removing press releases and events, leaving only Blogs and General publications. Finally, in popular technologies, omit all but top ranked new sources.
Above creates below MRI network.
4547 articles is near half of the maximum 10K articles allowed..and there are over a dozen additional neurotechnology keywords. Deal with it by making a series of small, focused networks that will later be merged together.
One for each technology. We have brain computer interfaces (BCI)s and cognitive control devices; imaging techniques like DTI, MRI, fMRI, EEG, MEG and electron microscopy; neuroprosthetics, optogenetics, microfluidics, and circuit modulation.
19 small networks merge into one ecosystem-level representation.
At First Glance
I was here, exploring the MRI subnetwork when I began writing this post, about half an hour ago. I thought “Maybe I would get better at navigating such networks if I documented how I explore one.”And so I have. Now I'm up to present and shall continue.
First step is communities.
Adding tag MRI. Noting brain function dominates conversations. NFL, learning disabilities.
Exploring “Cognitive Control” community, bottom right in purple. Impact of video games on the brain; Adam Gazzaley
A crazy community on reading the minds of patients in vegetative states reveals research showing that sometimes those who are unable to communicate are still conscious enough to understand and respond with their minds.
Links! Note links. Ok. Stem cell research is improving.
Moving to the Memory category. In mice, scientists can now erase bad memories and turn them into happy ones.
Hm. Let’s dig deeper. Add “memory” tag to these nodes. Turn on highlight mode to see where community sits relative to whole network.
Relayout and color by network centrality and we see a conversation dominated by a remarkable finding from MIT researchers that emotional connections of memories can be reversed.
Red nodes are most central. Turns out this research is conducted in the lab down the hall from me. I hadn’t even heard about it.
Back to the whole network to further explore dozens of communities, the largest of which consists of a mere 6% of the network.
A curious orange community considers how the brain controls body and behavior. Articles range from researchers connecting their brains together online (a scientist in Washington caused a colleague across campus to move his finger). The community is highly interspersed through the entire network and articles are broad.
Dig deeper with a custom layout. Build a new network out of what once was a community.
Recolor Network Based on Centrality
Controlling another person’s fingers is just the start. A 23 year old quadriplegic man moved his arm thanks to a neuroprosthetic implant. Apparently now you can fly a quadcopter with your mind, a specific region of the brain has been found to harbor neurons active when humans recognize social norms.
Here’s an interesting one. A community based on how men and women’s brains are wired differently. As I dig deeper, I find the results of a PNAS publication are questioned, resulting in a news bubble.
Let’s see what DARPA is up to.
Relayout the network to find discussions on DARPA’s new $40M BRAIN Initiative collaborative project that aims to create an implantable device that restores memories for soldiers with PTSD.
My brain is getting tired. Before I go to sleep here’s a community label graph to ponder. More soon.
Wow and we haven’t even gotten to additional layouts. So much more to explore! Amy