I’ve been thinking about dentistry less as a set of procedures and more as a problem of clinical vocabulary.
A patient does not walk in as a diagnosis. They walk in as a state: a broken tooth, bleeding gums, pain, esthetic dissatisfaction, medical fragility. The clinician’s first job is to infer what hidden process generated that state: caries, bruxism, periodontal breakdown, trauma, systemic disease, behavior, or some combination.
Then we ask the real healthcare question:
What future does this disease create if left alone, and what future does our intervention create instead?
The difference between those two futures is the beating heart of treatment.
But there is a limit. A healthcare team can only hold so much information in its head. A healthy patient with a small chip may need a simple model. A medically fragile patient near death could require nearly infinite detail to predict accurately — so the correct clinical move may be not “model harder,” but “recognize that this exceeds the dental vocabulary and route appropriately.”
This is where I think ideas from category theory, Poly, MDL, and even Gödel, Escher, Bach become surprisingly practical.
Clinical terms like “perio,” “caries,” “bruxism,” or “medically compromised” are not fixed truths. They are chunks — compressed vocabularies that help us act. Sometimes they are enough. Sometimes they must unfold into deeper medical, behavioral, or systemic grammars.
The art is choosing the smallest vocabulary that still captures the dangerous variation.
Too simple, and we miss risk.
Too complex, and the team drowns.
Maybe the future of clinical systems is not just better data. Maybe it is better compression.