Documentation Types
Rubric supports evaluation of various clinical documentation formats:
SOAP Notes
The standard format for outpatient encounters:
{
"soap_note": {
"subjective": "...", # Patient-reported symptoms, history
"objective": "...", # Exam findings, vitals, labs
"assessment": "...", # Clinical impression, diagnoses
"plan": "..." # Treatment, follow-up
}
}
H&P (History & Physical)
Comprehensive documentation for admissions:
{
"hp_note": {
"chief_complaint": "...",
"hpi": "...",
"past_medical_history": "...",
"medications": [...],
"allergies": [...],
"family_history": "...",
"social_history": "...",
"review_of_systems": {...},
"physical_exam": {...},
"assessment_and_plan": "..."
}
}
Procedure Notes
Documentation of procedures performed:
{
"procedure_note": {
"procedure": "...",
"indication": "...",
"consent": "...",
"technique": "...",
"findings": "...",
"specimens": [...],
"complications": "...",
"disposition": "..."
}
}
Clinical Note Evaluators
Completeness Evaluator
Checks for required documentation elements:
{
"type": "note_completeness",
"config": {
"note_type": "soap",
"required_sections": ["subjective", "objective", "assessment", "plan"],
"required_elements": {
"subjective": ["chief_complaint", "hpi", "ros"],
"objective": ["vitals", "exam"],
"assessment": ["diagnoses"],
"plan": ["treatment", "follow_up"]
},
"specialty_requirements": "internal_medicine"
}
}
Accuracy Evaluator
Validates factual correctness against source material:
{
"type": "note_accuracy",
"config": {
"source_types": ["transcript", "prior_notes", "labs"],
"check_medications": True,
"check_allergies": True,
"check_diagnoses": True,
"check_vitals": True,
"strict_mode": False # Allow reasonable inference
}
}
Hallucination Detection
Identifies claims not supported by source:
{
"type": "hallucination_detection",
"config": {
"sensitivity": "high", # low, medium, high
"categories": [
"unsupported_symptom",
"fabricated_history",
"incorrect_medication",
"wrong_patient_data",
"invented_lab_value"
]
}
}
Medical Coding Accuracy
Evaluates ICD-10 and CPT code suggestions:
{
"type": "coding_accuracy",
"config": {
"code_types": ["icd10", "cpt"],
"check_specificity": True, # e.g., E11.65 vs E11.9
"check_medical_necessity": True,
"check_documentation_support": True
}
}
Quality Metrics
SOAP Note Scoring
| Component | Weight | Criteria |
|---|
| Subjective | 25% | Complete HPI, relevant ROS, accurate history |
| Objective | 25% | Pertinent exam, accurate vitals/labs |
| Assessment | 25% | Appropriate diagnoses, clinical reasoning |
| Plan | 25% | Complete, addresses each diagnosis, follow-up |
Documentation Quality Score
Composite metric considering:
Quality Score = (Completeness × 0.3) +
(Accuracy × 0.4) +
(Coding × 0.2) +
(Readability × 0.1)
From Ambient Transcription
client.notes.log(
project="ambient-scribe",
# Raw audio/transcript from encounter
input_transcript=[
{"speaker": "provider", "text": "What brings you in today?"},
{"speaker": "patient", "text": "I've had this cough for about two weeks."},
# ...
],
# Optional: structured EHR data available during generation
context={
"prior_notes": [...],
"medications": [...],
"allergies": [...],
"labs": [...]
},
# AI-generated output
output={
"soap_note": {...},
"icd_codes": [...],
"cpt_codes": [...]
}
)
From Prior Note Summarization
client.notes.log(
project="chart-summarizer",
# Source documents
input_notes=[
{"date": "2024-01-15", "type": "progress_note", "text": "..."},
{"date": "2024-01-10", "type": "lab_result", "text": "..."},
# ...
],
# Generated summary
output={
"summary": "...",
"problem_list": [...],
"active_medications": [...],
"recent_labs": {...}
}
)
Copy-Forward Detection
Identify inappropriate duplication from prior notes:
{
"type": "copy_forward_detection",
"config": {
"similarity_threshold": 0.85,
"check_dynamic_fields": True, # Vitals, labs should change
"exempt_fields": ["allergies", "past_medical_history"],
"flag_unchanged_exam": True
}
}
Copy-forward of dynamic data (vitals, exam findings) without verification is a significant compliance and patient safety risk.
Specialty-Specific Requirements
Rubric supports documentation requirements by specialty:
| Specialty | Required Elements |
|---|
| Primary Care | Preventive care reminders, chronic condition management |
| Cardiology | Risk factor assessment, anticoagulation documentation |
| Psychiatry | Mental status exam, safety assessment, substance use |
| Surgery | Informed consent, procedure details, complication documentation |
{
"type": "specialty_requirements",
"config": {
"specialty": "psychiatry",
"required_elements": [
"mental_status_exam",
"suicide_risk_assessment",
"substance_use_screen",
"medication_reconciliation"
]
}
}
Next Steps
