ACBP logo

ACBP — SQL-Native Decision Spaces for Clinical Ops

Deterministic rules in PostgreSQL with verified theorems and interactive latencies.

Poster — Clinical Operations Updated: 2025-08-18 (Asia/Riyadh) Theorems Equation Paper (PDF)

Idea

ACBP(F,c) := ∧r∈R r(F,c). Flags F∈{0,1}^B, categories c∈C. We materialize the decision space D and strict valid masks M = πF(D) as SQL views/matviews.

Slogan. logic @ the speed of thought — deterministic, auditable, SQL-native.

Verified theorems

  • Soundness & completeness: D_sql = D (0 violations).
  • Projection: M_sql = πF(D_sql) (strict).
  • Present-only monotonicity: widening C' only grows D_present.
  • Deterministic δ: unique action per (F,c) when enabled.

Results (from DB)

ModelP50P95n
Clinic~818 ms~937 ms1,680
Inpatient~614 ms~750 ms1,680

Daily SLO: Clinic ≥95% under 920 ms; Inpatient ~83–91% under 700 ms (Wilson 95% lower bounds from app).

Why it works

  • Deterministic rule core: rules compile to SQL predicates.
  • Materialized decision space: fast joins, simple reasoning.
  • Verifier: SQL script checks soundness/coverage/dedup on DB.

Where it fits

  • OPD scheduling, no-show logic, clinic routing.
  • Inpatient discharge/transfer policies with audit trails.

Links