Where our analysis comes from
The source hierarchy our analysis follows — peer-reviewed pharmacology first, regulators as legal context.
Regimens™ evaluates everything you take — medications, supplements, herbs, foods, peptides, recreational substances — and flags interactions, contraindications, timing conflicts, and depletion patterns. The quality of that analysis is only as good as its sources.
This page exists because most AI wellness tools won’t tell you where their information comes from, and we think you deserve to know.
Our principle: science first, regulation as context
When peer-reviewed pharmacology research and a national regulator’s position diverge, we lead with the science.
National regulators like the FDA, DEA, MHRA, and EMA serve essential functions — they manage legal status, approve indications, and protect public safety. But regulatory frameworks lag the scientific literature, sometimes by decades. The 2002 FDA consumer advisory on kava, for example, persists in the regulatory record despite subsequent research substantially reframing those original case reports — and despite the 2020 Codex Alimentarius standard that recognizes traditionally prepared aqueous kava as a food.
When MDMA-assisted therapy showed Phase 3 efficacy for PTSD in 2021, the DEA’s Schedule I designation didn’t move. The science had advanced; the regulatory framework hadn’t.
Citing regulatory bodies as the primary scientific authority would mean accepting that latency as truth. We don’t.
Instead, our analysis follows a tiered hierarchy that prioritizes the underlying evidence and treats regulatory positions as legal-status context.
The source hierarchy
The tiers below are the universe of evidence the analyzer is built to reason within. Tier 1 leads; each lower tier is cited for what it is authoritative on, not as a substitute for stronger evidence. National regulators close the hierarchy at Tier 6 as legal context, not scientific authority.
Tier 1 — Peer-reviewed pharmacology and clinical research
This is where the science actually lives. Our analysis prioritizes:
PubMed-indexed primary research
Specifically randomized controlled trials, systematic reviews, and meta-analyses from journals indexed in MEDLINE.
Cochrane systematic reviews
Cochrane is the gold standard for evidence synthesis in medicine — independent, methodologically rigorous, and updated continuously. The Cochrane Complementary Medicine field publishes systematic reviews on integrative and complementary modalities under the same methodological standards.
Established pharmacology references
Goodman & Gilman’s The Pharmacological Basis of Therapeutics, Stockley’s Drug Interactions, and other clinical references that practicing pharmacologists actually use.
Preprint servers (bioRxiv, medRxiv)
When relevant emerging research hasn't yet completed peer review, we reference it explicitly as preliminary — not as established consensus.
DrugBank
A comprehensive online drug knowledgebase maintained by the University of Alberta and the Metabolomics Innovation Centre, combining pharmacological, chemical, and target data with peer-reviewed releases through Nucleic Acids Research. The referential drug data source for PharmGKB, KEGG, PubChem, and other major databases.
IUPHAR/BPS Guide to Pharmacology
Joint resource of the International Union of Basic and Clinical Pharmacology and the British Pharmacological Society. Peer-reviewed overviews of drug targets, ligands, and receptor pharmacology — the authoritative target-level reference for mechanism claims.
Within the peer-reviewed literature, certain journals serve as primary references for specific domains. Citations in your analysis frequently draw from:
Clinical and pharmacology
New England Journal of Medicine
The Lancet
JAMA (Journal of the American Medical Association)
BMJ (British Medical Journal)
Annals of Internal Medicine
Clinical Pharmacology & Therapeutics
British Journal of Clinical Pharmacology
Annals of Pharmacotherapy
Phytotherapy and natural products
Journal of Ethnopharmacology
Phytomedicine
Planta Medica
Phytotherapy Research
Journal of Natural Products
Fitoterapia
Nutrition
American Journal of Clinical Nutrition
Journal of Nutrition
Advances in Nutrition
British Journal of Nutrition
Regional and traditional medicine research
Chinese Medical Journal
Journal of Ayurveda and Integrative Medicine
Journal of Traditional and Complementary Medicine
Tier 1.5 — Clinical practice consensus
Where specialty-clinician consensus runs ahead of formal RCT evidence — peptide therapy, hormone optimization, specialty pharmacology — the practice literature documents the consensus position rigorously. Distinct from Tier 4: T1.5 is direct specialty-clinician consensus in fields where RCTs are structurally limited; T4 is third-party synthesis evaluating evidence with conventional clinical-evidence frameworks.
Peptide protocol literature
Specialty clinical reference texts and society consensus documents covering peptide therapy protocols where regulatory and economic barriers structurally limit RCT evidence.
Specialty-society practice guidelines
Clinical practice guidelines from specialty societies (e.g., Endocrine Society, NAMS, AACE) documenting consensus positions where practice-based knowledge precedes or substitutes for formal RCT evidence.
Specialty pharmacology consensus
Consensus documents from specialty clinical pharmacists addressing dosing, compounding, and protocol questions outside the standard pharmacopoeia.
Practice-specialty reference texts
Reference works documenting specialty practice consensus for compounds and protocols where conventional pharmacology references don't extend.
Tier 2 — Global scientific consensus bodies
Synthesis bodies that aggregate Tier 1 research into globally-coordinated positions.
WHO Expert Committees
Specifically the Expert Committee on Drug Dependence (controlled substances), the Expert Committee on the Selection and Use of Essential Medicines, the Expert Committee on Specifications for Pharmaceutical Preparations, and the Expert Committee on Biological Standardization.
Codex Alimentarius (FAO/WHO)
The international food standards body. Codex CXS 336R-2020, for example, is the current standard for kava as a food; we cite it where relevant.
NIH Office of Dietary Supplements (ODS)
For supplement safety, dosing, and interaction information. NIH ODS publishes fact sheets that synthesize the peer-reviewed literature on common supplements.
Endocrine Society Clinical Practice Guidelines
For hormone- and peptide-related compounds.
WHO Traditional Medicine Strategy
The WHO framework for evaluating traditional medicine systems against scientific and quality standards. Distinct from but informing the WHO Monographs program.
National Academy of Medicine (NAM)
Formerly the Institute of Medicine. Authoritative US synthesis body for clinical and population-health evidence; non-regulatory.
Dietary Reference Intakes (Food and Nutrition Board, National Academies)
The reference standard for nutrient intake recommendations across North America. Updated by the National Academies' Food and Nutrition Board.
MedlinePlus (NLM)
Consumer-health synthesis from the National Library of Medicine, drawing from peer-reviewed primary sources. Useful for accessible synthesis of established findings.
Linus Pauling Institute Micronutrient Information Center
Academic synthesis on vitamins, minerals, and dietary phytochemicals from Oregon State University. Peer-reviewed, free public access — the most rigorous open-access micronutrient reference.
European Food Safety Authority (EFSA)
The EU's scientific advisory body for food and feed safety. Issues opinions on novel foods, food additives, contaminants, and nutrient/health claims.
USDA FoodData Central
The USDA's authoritative food composition database, public domain. Foundation for most food-composition research globally; the food-compound equivalent of what DrugBank is for drugs.
Phenol-Explorer
Comprehensive polyphenol content database maintained by INRAE (France). The standard reference for polyphenol concentrations in foods — directly relevant to plant-compound claims.
EuroFIR eBASIS (Bioactive Substances in Food Information System)
The European bioactive-compound database under the European Food Information Resource consortium. Complements USDA FoodData Central with broader bioactive coverage.
USP Herbal Medicines Compendium and Dietary Supplement Verification Program
United States Pharmacopeia's standards for herbal product and dietary supplement identity, purity, strength, and quality. The pharmacopeial reference for supplement integrity.
Tier 3 — Specialized pharmacology references
For specific claim types where dedicated reference works are the standard:
Drug-induced nutrient depletion
Pelton’s Drug-Induced Nutrient Depletion Handbook, Cass’s Supplement Your Prescription, and the broader integrative medicine literature on medication-nutrient interactions.
Pharmacology mechanism
Goodman & Gilman as the primary reference; Katzung’s Basic & Clinical Pharmacology as secondary.
Drug interactions
Stockley’s Drug Interactions (Pharmaceutical Press), the Hansten and Horn Drug Interactions Analysis & Management reference.
Lactation safety
The NLM LactMed database (peer-reviewed, regularly updated).
Herb-drug interactions
Stockley’s Herbal Medicines Interactions (Williamson, Driver, Baxter, Pharmaceutical Press) — companion volume to Stockley’s Drug Interactions, the primary reference for herb-drug interaction claims.
Pharmacogenomics
PharmGKB (Stanford / NIH-funded Pharmacogenomics Knowledgebase) and the Clinical Pharmacogenetics Implementation Consortium (CPIC), which publishes peer-reviewed gene-drug prescribing guidelines in Clinical Pharmacology & Therapeutics. Cited where pharmacogenomic context affects drug response.
Clinical reference
Merck Manual Professional Edition (MSD Manual) — broad clinical reference for differential diagnostic and therapeutic context. Freely accessible.
Tier 4 — Integrative medicine evidence synthesis
Bodies that aggregate peer-reviewed evidence on integrative and complementary modalities, evaluating them with conventional evidentiary standards. Distinct from traditional medicine literature in Tier 5: integrative synthesis bodies use modern clinical-evidence frameworks; traditional pharmacopoeias preserve evidence in a different epistemic register.
National Center for Complementary & Integrative Health (NCCIH, NIH)
The NIH agency funding and synthesizing research on complementary and integrative health practices. Publishes evidence summaries, fact sheets, and clinical research findings.
Memorial Sloan Kettering Integrative Medicine — About Herbs
MSK's clinician-facing herb database, peer-reviewed by the institution's oncology and pharmacy services. Authoritative for herb-drug interaction concerns in oncology and clinical care contexts.
Natural Medicines Comprehensive Database
The Therapeutic Research Center's clinician-grade evidence database for natural products. Widely used by pharmacists and integrative practitioners for evidence-graded efficacy and interaction information.
European Scientific Cooperative on Phytotherapy (ESCOP) Monographs
Sixty-plus monographs harmonizing phytotherapy standards across Europe. Unlike Commission E, ESCOP monographs include explicit references — the European reference for evidence-based phytotherapy.
American Herbal Pharmacopoeia (AHP) Monographs and Therapeutic Compendia
Peer-reviewed by an international multi-disciplinary collective (botanists, chemists, herbalists, pharmacists, pharmacologists, physicians, toxicologists). Each monograph runs roughly thirty pages on a single botanical, integrating quality control, pharmacology, and therapeutic use.
WHO Monographs on Selected Medicinal Plants
World Health Organization global synthesis of clinical and traditional evidence on plant medicines. Each monograph applies modern evidentiary frameworks — clinical pharmacology, controlled trial summaries, contraindications, adverse effects — to traditional materia medica. Distinct from the WHO TCM Monographs in Tier 5 (which focus specifically on Traditional Chinese Medicine in the traditional register).
American Botanical Council / HerbalGram
Peer-reviewed quarterly journal and reference publisher for evidence-based herbalism. Publishers of the English translation of the German Commission E monographs.
Herbal Medicine: Expanded Commission E Monographs
Blumenthal et al., American Botanical Council (2000) — the English translation and expansion of the German Federal Health Authority's Commission E review of 324 herbs (1978–1994). Historical foundation for Western evidence-based phytotherapy.
British Herbal Pharmacopoeia and British Herbal Compendium
British Herbal Medical Association — quality standards for 169 herbs (BHP) plus volumes of scientific information and monographs (Compendium). The British counterpart to AHP.
Botanical Adulterants Prevention Program
Joint program of the American Herbal Pharmacopoeia, American Botanical Council, and the National Center for Natural Products Research at the University of Mississippi. Documents and addresses adulteration in commercial herbal products — directly relevant to "the supplement may not contain what the label says" claims.
Tier 5 — Traditional medicine literature
For compounds rooted in traditional medicine systems, we reference the relevant traditional pharmacopoeia explicitly as traditional-use evidence — distinct from RCT-grade modern research:
Ayurvedic Pharmacopoeia of India
For compounds in the Ayurvedic tradition.
WHO TCM Monographs
For compounds in Traditional Chinese Medicine.
EMA Herbal Monographs (Traditional Use)
These explicitly distinguish “traditional use” registration from “well-established use” registration. We preserve that distinction in our analysis.
WHO International Standard Terminologies on Traditional Medicine
The WHO standardization framework for traditional medicine terminology, supporting consistent classification of traditional-use evidence across WHO member states.
Pharmacopoeia of the People's Republic of China (PPRC)
The official pharmacopoeia of the PRC, including extensive monographs on Traditional Chinese Medicine materia medica with quality and identity standards.
These sources matter because dismissing them would mean ignoring documented evidence on compounds humans have used for centuries. Including them honestly means flagging them as the type of evidence they are — distinct from RCT-grade modern research, distinct again from integrative-medicine synthesis (Tier 4), preserved as their own knowledge tradition.
Tier 6 — National regulators (legal context)
National regulators close the source hierarchy because their authority is legal, not scientific. We cite them for what they are authoritative on: legal status, approved indications, formal warnings, and prescribing information for specific markets. When the regulatory position aligns with the current science, we say so. When it diverges, we name the divergence and present the science.
FDA prescribing information
For what's approved, what's contraindicated, and what's specifically warned against in US-marketed prescription drugs.
EMA (European Medicines Agency)
For European market approvals, herbal monographs, and post-market safety actions.
DEA scheduling
For US controlled substance status, framed clearly as legal status (not as scientific assessment of risk).
Other national regulators
MHRA (United Kingdom Medicines and Healthcare products Regulatory Agency), TGA (Australia Therapeutic Goods Administration), Health Canada (including the Drug Product Database and Natural Health Products Database), PMDA (Japan Pharmaceuticals and Medical Devices Agency), and Swissmedic. Cited where users' regional context calls for them.
DailyMed (NLM)
Structured FDA-derived drug labeling, maintained by the National Library of Medicine. Primary US source for current prescribing information in machine-readable form.
FDA Adverse Event Reporting System (FAERS)
FDA's pharmacovigilance database for post-market adverse event reports. Cited for known safety signals where applicable.
FDA pharmacogenomic biomarkers list
The FDA's regulatory list of pharmacogenomic biomarkers in approved drug labeling. Complements PharmGKB / CPIC (Tier 3) by reflecting where pharmacogenomic information has reached US regulatory recognition.
We frame these citations as what they are: regulatory positions, not scientific consensus. When the regulatory position aligns with current science, we say so. When it diverges, we explain the divergence and present the science.
What the corpus actually contains
The tier hierarchy above isn’t a list of sources the analyzer looks up at runtime. It’s the editorial universe of evidence Regimens is built to reason within — the texts, databases, references, and clinical materials that informed the analyzer’s training and that anchor the curated framing for specific compounds. When the corpus expands, the framing expands; the analyzer doesn’t gain new runtime reach, it gains better-grounded priors and more curated overrides.
How we read the sources
Here is the honest mechanical truth about how Regimens produces its analyses, because the editorial discipline depends on you understanding it.
Regimens uses Anthropic’s Claude as its analytical engine. Claude does not look up documents at runtime. It generates its responses from statistical patterns absorbed during training — a training corpus that included peer-reviewed pharmacology literature, regulatory advisories, structured clinical references, and the other source types named in the tier hierarchy above. So when Claude produces a claim about a drug interaction, that claim is genuinely informed by those sources. But it is not extracted from any one of them.
This matters because it shapes what “source fidelity” honestly means here. If we asked Claude to attach a specific paper to each individual claim, it would generate a plausible-sounding attribution that it did not actually consult — citation theater that looks rigorous but introduces fabrication into the very layer that’s supposed to defend against it. So we don’t do that.
What we do instead is the discipline this whole page describes:
We don’t fill gaps with plausibility. When the curated source hierarchy doesn’t support a claim, the analyzer is instructed to label the evidence as insufficient — not to generate a plausible-sounding answer to fill the silence.
We don’t smooth disagreement. When peer-reviewed pharmacology and a regulator’s position diverge, when traditional use and modern evidence disagree, when two clinical studies contradict each other — the analyzer is instructed to name the divergence, not paper over it.
We use the tier signal as the load-bearing source-fidelity output. Every finding in your report carries a tier label honestly indicating the strength of evidence behind it. T1 means peer-reviewed pharmacology and clinical research supports the claim. T6 means it’s grounded only in regulatory framing, not science. NONE means the corpus doesn’t support the claim at all and we’re saying so.
We curate context overrides for the compounds where general training is known to be weak. Where the analyzer would otherwise reason from stale or contested training data, Regimens injects curator-authored framing directly into the prompt so the analysis reasons from the curated framing instead.
This is what source-fidelity actually looks like at the layer Regimens operates. The tier hierarchy above is the universe of evidence the analyzer is built to reason within. The per-finding tier label tells you how strongly that universe supports each specific claim. That’s the honest commitment, and it’s what we’re building everything else on.
What we don’t do
We don't pretend our analysis is medical advice.
It isn't. Always consult your doctor and healthcare team before making changes to your regimen.
We don't claim diagnostic capability.
Regimens identifies documented interactions and patterns; it doesn't diagnose conditions or replace clinical evaluation.
We don't have access to your medical records.
Our analysis is based on the regimen you log, the personal health profile you share with us, and our source corpus. We can't account for medical history we don't know about.
We don't run live searches.
Regimens AI uses a frozen knowledge corpus, plus structured prompts that prioritize the source hierarchy above. New research published after the model's training date won't be reflected — and we say that directly when it might matter.
We're not HIPAA-covered.
Regimens is a working tool for personal pharmacology, explicitly not a clinical service or HIPAA-covered application. Our privacy policy explains how we handle the data you share with us.
How we work against AI’s failure modes
Our analysis is generated by Regimens AI, configured with the source hierarchy described above. Like all AI systems, it can occasionally generate plausible-sounding but inaccurate information.
Generative AI fabricates plausible-sounding sources when asked to attach specific citations to individual claims. Regimens doesn’t pretend otherwise. The defenses described on this page are designed to honor what the underlying mechanism actually does — and to refuse the kinds of attribution it can’t honestly produce. The four disciplines this page describes — a curated source corpus, a per-finding evidence-tier signal (T1, T1.5, T2, T3, T4, T5, T6, T0, NONE), curated context overrides for compounds where training data is thin or stale, and explicit gap-flagging — are what source fidelity actually looks like at this layer.
We don’t claim our analysis is perfect. We claim it’s better-grounded and more honestly framed than any wellness AI tool we know of, and we welcome corrections from clinicians, researchers, and informed users.
Questions or corrections?
If you believe a claim is misframed against the evidence, or a source in our corpus is wrong, we want to know. Email support@curiouswellnessgroup.com with specifics. We’ll investigate and update.