This work develops the mathematical machinery underneath the lab's classification research: neighborhood operators defined on a Hilbert space of clinical concepts, calibrated against observed data and self-enriching, meaning the operators refine their own neighborhoods as they classify.
It is the most formal statement of how the lab treats meaning as geometry, and it connects the applied classification work to the geometric program in Data Field Theory.
This is a technical report on internal research.
Related published work: Data Field Theory, a geometric framework for learning on Riemannian manifolds.