Inside the project
Prepared with the project’s development agent from source and retained evidence. This is a project-owner assessment, not an independent customer review. Dated checks were not rerun for this website.
NeuroPharmacology is an offline drug-interaction reasoning engine: enter a medication list and get structured, source-tagged what-if reasoning about shared metabolic enzymes, overlapping pharmacodynamic loads (QT, serotonergic, bleeding, sedation) and converging molecular targets. It carries ~23,600 drugs for identity/search and reasons deeply over ~6,000 where real data exists, with real FDA label text folded in as quote-bounded evidence. A wording guard keeps every output research-only — never prescribing advice. This is the family's highest-confidence module; it was verified against source this session.
What stands out
- Two honest tiers stated plainly: 23,599 drugs identity/search vs ~5,974 deep-reasoning
- Fully offline; live RxNav/ChEMBL/ClinicalTrials.gov only annotate
- Real openFDA/DailyMed label sections (89 drugs) as quote-bounded evidence
- Mechanism-first categorised hypotheses with counter-evidence, not a single severity word
- Enforced non-diagnostic wording
Current limits
- Deep pharmacology data only exists for a few thousand characterised drugs; the long tail stays identity-only (honest ceiling)
- ChEMBL coverage is occasionally idiosyncratic, so it annotates and never overrides the curated seed
- Non-diagnostic; no personalised dosing
Features & evidence
Identity index (23,599 drugs)
component-testedRxNorm ingredients/brands/combinations + WHO ATC class for search/normalisation.
Recorded evidence
staged + loaded (JSON snapshot)
Curated deep seed + CombinationReasoner
component-tested~97 hand-modelled drugs; 7 separated hypothesis categories.
Recorded evidence
reasoning tests incl. sertraline+tramadol gate
Offline ChEMBL enrichment (5,974 drugs)
component-testedTargets/enzymes/PD-loads for the characterised universe, no runtime network.
Recorded evidence
full macOS build succeeded this session, 2026-10
openFDA/DailyMed label layer
component-testedVerbatim boxed warning/contraindications/interactions/etc. as evidence.
Recorded evidence
89 drugs staged 2026-10 (54 boxed warnings, 84 contraindications, 87 warnings)
Review updates & corrections
- No material change since 17 Oct; PharmacologyLabelStore.swift verified intact this session
Workflow walkthrough
A guide to the native workflow. These steps describe the app; only the recordings and images embedded on this page are available as website demonstrations.
- Enter or import a medication list
- Each drug is normalised against the 23.6k identity index
- The reasoner returns categorised hypotheses (shared CYP pathways, QT/serotonergic/bleeding load overlap, converging targets)
- Real FDA label snippets attach as evidence
- Everything is framed as research, never a recommendation
Planned captures · not yet available
- Combine / What-If panel with drug-count metrics