qp — Quantitative Human Physiology

A tested, importable library of the reusable quantitative models from Feher's Quantitative Human Physiology, 3rd ed., built for CICM / ANZCA Primary Exam reasoning — spanning cardiovascular, respiratory, renal, membrane, endocrine and cross-cutting acid–base physiology (Henderson–Hasselbalch, anion gap, Stewart strong-ion difference). Every equation is an AtomicEquation object — a compute function with declared parameters, units, dependencies, and Feher provenance — not just a formula in prose. The value is not a bag of equations but an integrated, queryable network: seed a few bedside numbers and the graph fills in the rest.

367 canonical equations across 9 domains:

Domain	n	Domain	n	Domain	n
foundations	38	membrane	30	excitable	29
nervous	36	cardiovascular	58	respiratory	62
renal	42	gastrointestinal	41	endocrine	31

Two tiers

• qp/ — the full skill: the 367 equations, the reasoning layers, the beyond-Feher extensions/ tier, the reference docs, the test suites, and a single deterministic acceptance gate.
• qpm/ — a minimal runtime tier: the same 367 equations + reasoning layers, no tests/gate/references, plus a dependency-free integrity self-check.

Both tiers return identical values on shared equations.

Quickstart

from scripts.foundations.thermodynamics import nernst_equation
nernst_equation.compute(z=1, C_out=4, C_in=140)   # -95.0  (mV — output_units='mV')

# Integrative simulation: seed the bedside numbers, read the whole state
from scripts.eqgraph import build_graph
g = build_graph()
r = g.propagate(dict(HR=80, SV=65, EDV=120, ESV=55, Hb=14, S_O2=0.97,
                     P_O2=95, P_50=26.8, n=2.7, SBP=125, DBP=78))
f = r["fired"]                       # {equation_id: value}, 60+ entries
f["cardiac_output"], f["mean_arterial_pressure"]   # CO and MAP fall out

Reasoning layers (the network, not just the atlas)

• eqgraph — dependency DAG + propagate() forward-chaining integrative simulation, dimensional coupling and inference.
• primitives — ~7 universal kernels (log-ratio, saturation, first-order, linear-flux, …); 47 execution-proven memberships showing disparate equations (Nernst ≡ Henderson-Hasselbalch ≡ Gibbs; Michaelis-Menten ≡ Hill) are one form.
• feedback — 7 physiological feedback loops (baroreflex, HPA/HPT axes, TGF …) represented beside the acyclic DAG; the endocrine axes are formula-verified.
• dependency_types — AND/OR dependency typing: which inputs are jointly required vs substitutable alternatives (e.g. cardiac output by any of three methods).
• eqforms — every equation classified by mathematical form on two independent grounds.
• extensions/ — definitional relations beyond Feher (instrument response, thermoregulation heat balance, PK timing), a tier separate from the canonical 367.

Verify

cd qp  && python -m scripts.run_gate   # full gate: idempotence + all suites + docs + integration
cd qpm && python -m scripts.smoke      # minimal-tier integrity check

Grounding discipline

Every value is executable and anchored to an independent external reference, never a circular self-test. Additions are additive (no equation collapsed or deleted); anything that cannot be grounded without fabricating corpus is recorded as an evidenced rejection rather than stubbed. See qp/ROADMAP_LEDGER.md and qp/QPM_ASSESSMENT.md.

Orthogonal context

context/obsidian-voice-context.md is a separate, provenance-labelled retrieval artifact for an Obsidian voice plugin: definitions, schemas, diagrams, and equation-synthesis notes extracted from live teaching sessions and vault sources (e.g. the Stewart strong-ion acid-base approach). Each item carries its source anchor and a provenance class (extracted / inferred / assumed / session-only), with uncertainty preserved.

It is deliberately outside the canonical registry — it does not add to, merge with, or alter the 367 equations, and is never imported by qp/ or qpm/. Anything promoted from it must re-enter through qp's normal grounded authoring and acceptance gate.

Source: Joseph J. Feher, Quantitative Human Physiology: An Introduction, 3rd ed.