Cómo evaluamos el riesgo de alucinación de la IA clínica
La narrativa de evidencia de Notat no es un porcentaje mágico de precisión. Es una metodología transparente: extraer primero los hechos clínicos, generar notas a partir de esos hechos, mostrar el contexto en bruto al clínico y revisar cada resultado frente a la evidencia.
Método de evaluación
1. Build the reference facts
A reviewer reads the encounter material and creates a reference set of clinical facts: symptoms, negatives, medication changes, assessment, plan, safety-netting, and coding-relevant details.
2. Compare transcript-direct output
We generate a note from transcript-style input and mark unsupported statements, missing high-salience facts, incorrect attribution, and invented certainty.
3. Compare FactsContext output
We generate documentation from extracted facts and review whether each sentence is supported by a visible fact. The clinician-facing fact list is evaluated as part of the output, not hidden.
4. Record limits, not magic numbers
We do not publish fake accuracy percentages. Each evaluation note includes dataset scope, review date, known limitations, and examples of what the system still requires clinicians to verify.
Qué marcan los revisores
Unsupported clinical assertion
Wrong medication, dose, frequency, or route
Missing red-flag negative or safety-net advice
Wrong diagnosis certainty: possible vs established
Wrong speaker attribution
Unsupported ICD-10 suggestion
Clinician-visible evidence for each key statement
Facts reusable for notes, codes, referrals, and patient instructions
Nota interna fechada
Nota de evaluación actual, 2026-07-01: esta página documenta el método y los ejemplos cualitativos utilizados para evaluar la arquitectura FactsContext. No afirma validación externa ni precisión universal. El próximo hito de evidencia debería ser una revisión ciega y estratificada por especialidad, con el tamaño del conjunto de datos, el acuerdo entre revisores y la tasa de afirmaciones no respaldadas reportados abiertamente.
Extracción de hechos frente a generación directa desde la transcripción
Medication change discussed twice
Riesgo de la transcripción directa
“Increase amlodipine to 10 mg and stop lisinopril.” The transcript contained a correction: the clinician first considered stopping lisinopril, then decided to continue it after reviewing renal function.
Resultado de FactsContext
Facts: amlodipine increased to 10 mg daily; lisinopril continued; renal function normal; review in 6 weeks. Note generated from those facts only.
Por qué importa
Transcript-direct generation can smooth over corrections. FactsContext preserves the final decision as a discrete fact before writing.
Negative finding matters
Riesgo de la transcripción directa
“No neurological symptoms.” The actual encounter only documented no saddle anesthesia and no bladder symptoms; leg radiation was present.
Resultado de FactsContext
Facts: left leg radiation to calf; SLR positive left; no saddle anesthesia; no bladder or bowel symptoms. Note keeps the negatives specific.
Por qué importa
Broad negative statements are risky. The fact list keeps the clinical context granular and reviewable.
Code suggestion requires evidence
Riesgo de la transcripción directa
Suggested J44.1 for COPD exacerbation without showing the symptom or treatment evidence.
Resultado de FactsContext
Facts: increased breathlessness, purulent sputum, prednisolone burst, antibiotics started. Suggested J44.1 with those facts as evidence.
Por qué importa
The code is easier to verify because the reason for the suggestion is visible, not buried in prose.
Limitaciones conocidas
Clinician review remains mandatory. Notat drafts; clinicians verify and sign.
The method reduces unsupported statements by architecture, but no clinical AI should claim zero hallucinations.
Small internal evaluations are useful for engineering direction, not a substitute for external clinical validation.
Specialty, language, audio quality, speaker overlap, and local coding rules can change performance.
Lee la evidencia y luego revisa tus propios hechos.
El método de evaluación es simple porque el producto está diseñado para ser inspeccionable: primero los hechos, después la nota, siempre revisión clínica.
Prueba Notat gratis