A Measure of Physiological/ Performance Measures
28 Apr 2017Recently read this piece on Elon Musk’s most recent venture, Neuralink. Fascinating article for a fascinating company.
To set the stage for Musk’s vision for the company, Urban recaps not only the whole of human history but also the history/ evolution of cephalization generally. Implying that arguably we humans are the peak/ pinnacle of a long process of nervous system evolution, driven by the utility of real-time sensing (input), processing (computation) and reacting (output) for biological organisms.
This perspective sets the stage well for Neuralink’s ambitions. It also gives some evolutionary biological intuition for why computational/ information technologies have become so widespread: they are extensions of this cephalization process, results of evolutionary forces pushing for better cognition to serve better biological fitness.
A Measure of Physiological Measures
What interested me most in this article, however, was more practical and low-level (though I guess still fairly “meta”). In reviewing the current state of the art in brain-machine interfaces (BMIs), he uses a framework that I think could be applied more generally to all devices that aim to measure human physiological/ functional dynamics.
His evalution framework consists of 3 what-he-calls “broad criteria”: scale, resolution (spatial/ temporal) and invasiveness.
I think if these measures were made more specific (i.e. quantified) and taken seriously, they could be highly useful for evaluating potential of novel biometric technologies, as well as envisioning the ideal (arguably more important).
In other words, these 3 measures constitute a 3D meta-space on technologies, giving a sense/ intuition for which subspace a perfect technology would reside and further implying the directions we need to tend to improve the current state.
Blood Biometrics As An Example
For example, the first application of this evaluation space that came to mind was blood biometrics.1
For many reasons, there is a lot of interest in developing technology that provides a continuous measure of blood biomolecular (protein, genomic, etc.) composition.2 However, it’s difficult to imagine how this might be accomplished given all the layers of tissue between the most reasonable measure space (outside the body) and arteries/ veins (which permeate the innermost layers of cardiovasculatured organisms).
So it’s useful to use this evaluation criteria to ask the question: what is the current state of the art in blood biometrics? And what would the ideal blood biometric look like? That is, where in the space of scale, resolution (spatial/ temporal) and invasiveness are we, and where should we be going? And as you’ll see this brings us to all sorts of interesting questions.
What would be a decent scale for blood biometrics, that is volume of blood measured in a particular data point? That is, how much volume should we gather with each sample to select a portion representative of that individual’s whole blood?
This also leads naturally into thinking about spatial resolution, which for blood would be anatomical locations of sample collection: arms vs legs, arterial vs venous vs capillary. If we have it at one location, say a venous port in one of the arms, this would be of a smaller spatial resolution than if the samples were from multiple locations at each data point, say a venous port in an arm and a leg. (This would also probably increase scale as well, since in this case there is a relationship between spatial resolution and scale.)
Then we could also think about temporal resolution: how often do we have to/ should we sample to capture the most relevant physiological/ functional dynamics in blood? That is, what are the frequency ranges that capture the most interesting biomolecular dynamics in blood?
And invasiveness: are there ways in which we can measure biomolecular composition in blood without a port, without drawing blood outside the body? This differientiates approaches: to take blood out of the body would likely take a biochemical approach (some sort of immunological assay) whereas to measure inside without taking out would likely need a biophysical one (taking physical measurements that are representative of biomolecular composition, which would likely be very crude and of low resolution).
Biomedical Engineering And Science
We can see in all the above cases, the engineering considerations (the evalution criteria) actually give questions that may be more in the realm of science.
This shows the close relationship between biomedical engineering and biomedical science. The science isn’t robust/ quantitative/ “hard” enough to fully support the engineering discipline, so often engineering approaches on science.
And further, since biomedical science and engineering are fundamentally machinic (i.e. biological organisms are machines) then the science is more like the process of reverse engineering than anything else.
1: I use the term biometric very generally to mean any measure of biological process or function. Go Back
2: Blood is the slow-information highway for the body (as opposed to the fast-information highway of the nervous system). Blood facilitates travel/ transport for all the components of the endocrine and immune systems, which although both operate at lower frequencies/ slower speeds than the nervous system I think harbor much more interesting and exciting progressive dynamics in the long-term (i.e. learned immunity, physiological adaptation, etc.) Go Back