So, I have a few big announcements to make. But that’s not what this post is about.
This post is about a bit of a thunderclap moment I had. One that relates to the type of research lab I want to run 1.
I realized that I want three main pillars for my work.
NeuroTech
The main pillar is NeuroTech. This is fairly straightfoward - I want to have a rigorous 2 research program in neuromodulation - both DBS and BCI.
What may be a bit less clear is that I want to use AI as a means to effect a Control Theoretic Framework for Neuromodulation. That is, I’m not interested in building any whimsical AGI nonsense - I want to leverage tools that let humans take control over their own conditions. The tighter that control is, the better - I’d love to see people take direct control of their own health.
That means we’ll have to focus on concepts like distributed/network control, control barrier functions, optimal control in model predictive paradigms, etc. There’s a dearth there that I might fit neatly into given my “clinical reverse engineer” vibes.
MedTech
The secondary pillar is MedTech. This is a little less straightforward - I want to bring control theoretic rigor to what physicians have always done.
This is in direct opposition to efforts that take autonomy away from the patient-physician relationship. That is, I want to help bring medical care to the same level of respect-for-client that engineers have for their “customers”. Theirs is not a role to argue about values or goals - theirs is to lend their expertise so others can achieve the dreams they laid out for themselves 3.
This will focus on things like model-predictive control, on translating current (patho)physiologic models into mathematical modules, and formalizing optimal control into the health system. Part of this will likely have to be done in a startup/FRO structure.
HealthTech
The last pillar is HealthTech. Which seems like it’s covered above, but it’s the explicit complement.
This is all the factors of health that are outside of Medicine - which is actually most of this. Things like preventative care and socioeconomic justice fall here. This is so crucial, so important, to prevent parts of our society from being allowed to rot so a few folks can make a lot of money 4.
This will be a whole other thing that has a broader, humanistic scope that pulls from methodologies and frameworks developed above - maybe even some AI, as long as it is ethical and locally controllable. Part of this may have to leverage that LSAT I just took…
Which has changed significantly since I decided not to do residency. That’s a story for another day, but suffice it to say my adjusted aspirations for my research efforts are much more tightly aligned with my own historical disposition. ↩︎
Rigorous hear does not refer to reductionist empiricism - NHST is pretty garbage in the lab, it’s definitely garbage in the real world setting where you can’t keep confounds constant. So this rigor is more of an engineer’s rigor, one that prioritizes models and making quick decisions and iteratively improving alongside actual action. I think this is more rigorous for IRL inferences, but others will obviously consider this the opposite. This is not for them. ↩︎
This is in stark contrast with Tech - which seeks to impose values top-down with little regard for the customer or public. Much more on this… ↩︎
n-1 now. ↩︎