What MedExtract Does

Everything You Need to Extract Lab Data

Upload any Spanish lab report and get back coded results your systems already understand. Works with PDFs, scans, and phone photos — no manual data entry, no transcription errors.

Core

Any Report, Any Condition

Blurry phone photo? Faded 10-year-old scan? Clean PDF? Doesn't matter. Upload it, get structured results — over 99% accuracy on real documents.

Self-Hosted

No OpenAI. No Google. No Third Parties. Ever.

Every AI model is built and trained by our team. Your patient data never leaves your infrastructure — not even for a millisecond.

Universal Test Identification

Your lab says 'Glucosa basal.' Another says 'Glicemia en ayunas.' A third writes 'GLU.' MedExtract recognizes 27,000+ variants and maps each to the correct LOINC code.

Standards

FHIR R4 Output

Every response comes back as a valid FHIR R4 Bundle — ready to drop straight into your EHR or clinical system. No transformation layer needed.

Built for Spain, Not Translated

We know that a comma means decimals here, that 'hematíes' and 'eritrocitos' are the same thing, and that every comunidad autónoma formats reports differently.

Damaged Documents? No Problem.

Tilted scan? Low contrast fax? MedExtract straightens, sharpens, and cleans up documents automatically.

Catches What Humans Miss

Every result is checked against clinically valid ranges. If a glucose reading looks impossibly high or a hemoglobin is suspiciously low, we flag it before it reaches your system.

One Standard Unit System

mg/dL, mmol/L, g/L — labs use different units for the same test. MedExtract converts everything to international standard units so your data is always comparable.

Runs in Your Hospital, Scales With You

Deploy on your own servers. Auto-scaling, monitoring, and audit logs built in — your IT team gets full control without building anything.

Security

Zero Data Retention

Patient data flows through memory and disappears. Nothing is stored, cached, or written to disk. That's not a feature — it's the architecture.

Advanced

Multi-Layer AI for the Hardest Documents

When a standard read isn't enough, MedExtract activates additional AI layers to handle the toughest scans — handwritten notes, mixed layouts, or severely degraded originals.

Architecture

How We Get to 99%+ Accuracy

Three layers of AI work together. Each one catches what the previous layer might miss — so even the most damaged documents get accurate results.

Step 1 · Read

Document Analysis

We extract every piece of text and structure from your document — PDFs, scans, photos — regardless of format or quality.

Step 2 · Verify

Cross-Checking

Every extracted value is cross-checked for accuracy. Confidence scores tell you exactly how reliable each result is.

Step 3 · Rescue

Deep Analysis

For the hardest documents, an additional AI layer kicks in to recover results that simpler tools would miss entirely.

Everything runs within your infrastructure. Zero external API calls.

Standards

Plug Into Your Existing Systems

Results come back as valid FHIR R4 Bundles with LOINC codes and standard units — ready for your EHR, data warehouse, or analytics platform. No custom mapping required.

FHIR R4 Bundle your EHR can import directly
Every test coded to LOINC (27,000+ recognized)
Units converted to international standards automatically
Reference ranges and abnormal value flags included

FHIR R4 Bundle — response.json

{  "resourceType": "Bundle",  "type": "collection",  "timestamp": "2026-02-24T14:30:00Z",  "entry": [    {      "resource": {        "resourceType": "DiagnosticReport",        "id": "report-001",        "status": "final",        "category": [          {            "coding": [              {                "system": "http://terminology.hl7.org/CodeSystem/v2-0074",                "code": "LAB",                "display": "Laboratory"              }            ]          }        ],        "code": {          "coding": [            {              "system": "http://loinc.org",              "code": "24323-8",              "display": "Comprehensive metabolic 2000 panel"            }          ]        },        "effectiveDateTime": "2026-02-24T10:00:00Z",        "result": [          {            "reference": "Observation/glucose-001"          }        ]      }    },    {      "resource": {        "resourceType": "Observation",        "id": "glucose-001",        "status": "final",        "category": [          {            "coding": [              {                "system": "http://terminology.hl7.org/CodeSystem/observation-category",                "code": "laboratory"              }            ]          }        ],        "code": {          "coding": [            {              "system": "http://loinc.org",              "code": "2345-7",              "display": "Glucose [Mass/volume] in Serum or Plasma"            }          ]        },        "valueQuantity": {          "value": 95,          "unit": "mg/dL",          "system": "http://unitsofmeasure.org",          "code": "mg/dL"        },        "referenceRange": [          {            "low": {              "value": 70,              "unit": "mg/dL"            },            "high": {              "value": 100,              "unit": "mg/dL"            }          }        ]      }    }  ]}

See It Working With a Real Report

Try it with a sample report or bring your own. Full features, no credit card, no commitment.

Read the Docs Contact Us

Plug Into Your Existing Systems

Results come back as valid FHIR R4 Bundles with LOINC codes and standard units — ready for your EHR, data warehouse, or analytics platform. No custom mapping required.

FHIR R4 Bundle your EHR can import directly

Every test coded to LOINC (27,000+ recognized)

Units converted to international standards automatically

Reference ranges and abnormal value flags included

{ "resourceType": "Bundle", "type": "collection", "timestamp": "2026-02-24T14:30:00Z", "entry": [ { "resource": { "resourceType": "DiagnosticReport", "id": "report-001", "status": "final", "category": [ { "coding": [ { "system": "http://terminology.hl7.org/CodeSystem/v2-0074", "code": "LAB", "display": "Laboratory" } ] } ], "code": { "coding": [ { "system": "http://loinc.org", "code": "24323-8", "display": "Comprehensive metabolic 2000 panel" } ] }, "effectiveDateTime": "2026-02-24T10:00:00Z", "result": [ { "reference": "Observation/glucose-001" } ] } }, { "resource": { "resourceType": "Observation", "id": "glucose-001", "status": "final", "category": [ { "coding": [ { "system": "http://terminology.hl7.org/CodeSystem/observation-category", "code": "laboratory" } ] } ], "code": { "coding": [ { "system": "http://loinc.org", "code": "2345-7", "display": "Glucose [Mass/volume] in Serum or Plasma" } ] }, "valueQuantity": { "value": 95, "unit": "mg/dL", "system": "http://unitsofmeasure.org", "code": "mg/dL" }, "referenceRange": [ { "low": { "value": 70, "unit": "mg/dL" }, "high": { "value": 100, "unit": "mg/dL" } } ] } } ]}