> ## Documentation Index > Fetch the complete documentation index at: https://docs.medlistiq.com/llms.txt > Use this file to discover all available pages before exploring further. # Extract a deduplicated medication list from clinical PDFs > Accepts one or more clinical PDFs (referral packets, H&Ps, MARs, progress notes, discharge summaries) for a single patient and returns a curated, deduplicated medication list. Each medication carries the document / page / text-snippet evidence it was inferred from across all submitted documents. Every medication is normalized to RxNorm at the most-specific level the source supports (SCD / SBD / GPCK / BPCK when ingredient + strength + dose form are available; ingredient-level codes otherwise). Ingredients are surfaced with per-ingredient strengths from RxNorm, routes are coded in SNOMED CT + NCI Thesaurus, and combo products / branded packs collapse into one entry per drug. Limits: up to 15 files per request, 50 MB per file, 150 MB total, 200 pages combined. Larger packets should be split into multiple calls; server-side batching for very large packets is on the roadmap. ## OpenAPI ````yaml https://api.medlistiq.com/openapi.json post /v1/med-lists/from-documents openapi: 3.1.0 info: title: MedListIQ description: Infer a deduplicated medication list from FHIR medication resources. version: 0.1.0 servers: - url: https://api.medlistiq.com description: Production security: [] paths: /v1/med-lists/from-documents: post: summary: Extract a deduplicated medication list from clinical PDFs description: >- Accepts one or more clinical PDFs (referral packets, H&Ps, MARs, progress notes, discharge summaries) for a single patient and returns a curated, deduplicated medication list. Each medication carries the document / page / text-snippet evidence it was inferred from across all submitted documents. Every medication is normalized to RxNorm at the most-specific level the source supports (SCD / SBD / GPCK / BPCK when ingredient + strength + dose form are available; ingredient-level codes otherwise). Ingredients are surfaced with per-ingredient strengths from RxNorm, routes are coded in SNOMED CT + NCI Thesaurus, and combo products / branded packs collapse into one entry per drug. Limits: up to 15 files per request, 50 MB per file, 150 MB total, 200 pages combined. Larger packets should be split into multiple calls; server-side batching for very large packets is on the roadmap. operationId: from_documents_v1_med_lists_from_documents_post requestBody: content: multipart/form-data: schema: $ref: >- #/components/schemas/Body_from_documents_v1_med_lists_from_documents_post required: true responses: '200': description: Successful Response content: application/json: schema: $ref: '#/components/schemas/FromDocumentsResponse' '401': description: Missing or invalid API key. '422': description: >- Validation failure — no files, wrong content type, files too large, or total page count exceeds the synchronous limit (200 pages). '429': description: Rate limit or monthly cap exceeded. '503': description: >- OCR service is not configured or temporarily unavailable. Retry with backoff. security: - BearerAuth: [] components: schemas: Body_from_documents_v1_med_lists_from_documents_post: properties: files: items: type: string contentMediaType: application/octet-stream type: array title: Files external_patient_id: anyOf: - type: string - type: 'null' title: External Patient Id type: object required: - files title: Body_from_documents_v1_med_lists_from_documents_post FromDocumentsResponse: properties: medications: items: $ref: '#/components/schemas/ExtractedMedication' type: array title: Medications description: >- Reconciled medication list. Ordered by the section recency of each medication's canonical mention, then alphabetically by drug name. type: object required: - medications title: FromDocumentsResponse description: |- Body shape of ``POST /v1/med-lists/from-documents``. Body-only by design — request id, processing time, page counts, and ruleset version live in response headers (``x-request-id``, ``x-processing-time-ms``, ``x-document-count``, ``x-total-page-count``, ``x-output-medication-count``, ``x-ruleset-version``). ExtractedMedication: properties: drug_name: type: string title: Drug Name description: >- Human-readable name for the medication. When we resolved a specific RxNorm code (SCD / SBD / GPCK / BPCK), this is the catalog's canonical name for that code — already including ingredient, strength, and dose form (e.g. "penicillin V Potassium 250 MG Oral Tablet"). When we could only resolve to an ingredient-level concept, this is the lexicon-canonical name ("Metformin"). The structured fields below carry the same data in machine-readable form and reflect what the source document actually said — which may differ from the canonical strength when the patient is on a non-standard dose. examples: - penicillin V Potassium 250 MG Oral Tablet rxnorm_code: anyOf: - type: string - type: 'null' title: Rxnorm Code description: >- Most-specific RxNorm rxcui we could resolve for this medication. We prefer SCD / SBD / GPCK / BPCK (ingredient + strength + dose form) over less-specific term types, and only fall back to ingredient-level codes (IN / PIN) when the inputs lacked the fields we'd need to pin down a specific drug. Use rxnorm_tty to tell which level you got. Null when no rxcui matched at all (rare — usually uncommon drugs not in RxNorm). examples: - '834061' rxnorm_tty: anyOf: - type: string - type: 'null' title: Rxnorm Tty description: >- Term type of rxnorm_code. SCD / SBD / GPCK / BPCK indicate a fully-specified drug (use confidently for prescribing). SCDC / SBDC indicate ingredient + strength but no dose form. IN / PIN / MIN / BN indicate ingredient-only — treat as a fallback for class-level clinical decision support. Null when rxnorm_code is null. examples: - SCD ingredients: items: $ref: '#/components/schemas/Ingredient' type: array title: Ingredients description: >- Ingredients of this medication, each with its RxNorm code, term type, and per-ingredient strength when known. A single-ingredient drug has one entry; a multi-ingredient drug (e.g. amlodipine / valsartan) has one entry per ingredient. Use the rxnorm_code on each as the key for drug-interaction checks and class-level allergy matching, where the SCD-level top-level rxnorm_code is usually too specific. Empty when we could not resolve any ingredient. dose_quantity: anyOf: - type: string - type: 'null' title: Dose Quantity description: >- Numeric portion of the parsed dose, kept as a string so range values like '1-2' and decimals like '12.5' survive verbatim. Pair with dose_unit. Null when no dose was detected. examples: - '500' dose_unit: anyOf: - type: string - type: 'null' title: Dose Unit description: >- Unit portion of the parsed dose (e.g. 'mg', 'mcg', 'ml', 'tablets'). Lowercase. Null when no dose was detected. examples: - mg route: anyOf: - type: string - type: 'null' title: Route description: >- Normalized route abbreviation (PO, IV, IM, SC, INH, TOP, VAG, PR, TD, OPH, OTIC, NAS). Populated from the source text when one was mentioned; otherwise enriched from the resolved RxNorm dose form when possible. examples: - PO route_snomed_code: anyOf: - type: string - type: 'null' title: Route Snomed Code description: >- SNOMED CT code for the route, when ``route`` is set and matches a code in our route table (covers the ~30 most common administration routes — 95%+ of real use). ``http://snomed.info/sct`` is the implicit system. ``null`` when ``route`` is null or unrecognized. examples: - '26643006' route_ncit_code: anyOf: - type: string - type: 'null' title: Route Ncit Code description: >- NCI Thesaurus (NCIT) code for the route, paired with ``route_snomed_code``. The two coding systems target different downstream consumers (FHIR / clinical-decision support tend to expect SNOMED; oncology and research datasets tend to expect NCIT). ``http://ncithesaurus.nci.nih.gov`` is the implicit system. examples: - C38288 dose_form: anyOf: - type: string - type: 'null' title: Dose Form description: >- Dose form phrase (``oral tablet``, ``vaginal system``, ``transdermal patch``, …). Lowercase. Populated from the source text when present; otherwise enriched from the RxNorm canonical name when we resolved a specific RxCUI. ``null`` when we couldn't derive one. examples: - oral tablet frequency: anyOf: - type: string - type: 'null' title: Frequency description: >- Normalized frequency abbreviation (BID, TID, QHS, PRN, Q6H, etc). Phrases like 'twice daily' are mapped to BID. examples: - BID sources: items: $ref: '#/components/schemas/ExtractedMedicationSource' type: array title: Sources description: >- Per-mention evidence — one entry per detected mention across all input documents. type: object required: - drug_name - sources title: ExtractedMedication description: One reconciled medication, merged across all submitted documents. Ingredient: properties: rxnorm_code: type: string title: Rxnorm Code description: RxNorm rxcui for the ingredient. examples: - '83367' name: type: string title: Name description: Canonical ingredient name from RxNorm. examples: - atorvastatin rxnorm_tty: type: string title: Rxnorm Tty description: >- RxNorm term type — typically IN (Ingredient), PIN (Precise Ingredient — salt forms), or MIN (Multi-Ingredient). examples: - IN strength_quantity: anyOf: - type: string - type: 'null' title: Strength Quantity description: >- Numeric strength of this ingredient WITHIN the parent medication, as published by RxNorm (read from the ingredient's Semantic Clinical Drug Component). ``"800"`` for the 800 mg sulfamethoxazole component of Bactrim. Pair with ``strength_unit``. Null when the parent drug couldn't be resolved to a specific RxCUI or RxNorm doesn't publish a strength for this combination. examples: - '800' strength_unit: anyOf: - type: string - type: 'null' title: Strength Unit description: >- Unit for ``strength_quantity`` — e.g. ``"mg"``, ``"mcg"``, ``"mg/hr"`` (transdermal patches and vaginal systems). Lowercase. Null when ``strength_quantity`` is null. examples: - mg type: object required: - rxnorm_code - name - rxnorm_tty title: Ingredient description: |- One ingredient of a medication, with the strength RxNorm publishes for it within the parent product. Single-ingredient drugs surface one of these on the parent medication; multi-ingredient combinations (e.g. amlodipine / valsartan) surface one entry per ingredient. ExtractedMedicationSource: properties: document_name: type: string title: Document Name description: Filename of the input PDF this mention came from. examples: - discharge_summary.pdf page: type: integer minimum: 1 title: Page description: 1-indexed page number within the source PDF. examples: - 3 evidence_text: type: string title: Evidence Text description: The raw text line where the mention was matched. examples: - Metformin 500 mg twice daily type: object required: - document_name - page - evidence_text title: ExtractedMedicationSource description: |- One evidence span — where in the input PDFs we found this med. The reconciler aggregates one entry per detected mention across all submitted documents, so a med corroborated by 3 docs has 3 sources. securitySchemes: BearerAuth: type: http scheme: bearer bearerFormat: API Key description: >- MedListIQ API key (format: `ml_...`). Create one at [Dashboard → API Keys](https://medlistiq.com/dashboard/keys). ````