Predictive Document Intelligence Overview

Validate clinical documentation against payor and provider rules before prior authorization submission

What is Predictive Document Intelligence?

Predictive Document Intelligence evaluates clinical documentation before a prior authorization is submitted to a payor. It analyzes uploaded documents against validation rules to catch issues - missing signatures, outdated assessments, mismatched diagnosis codes - before they cause denials. Validation does not prevent submission as you can always proceed with creating and submitting a prior authorization, regardless of whether all rules pass.

Rules come from two sources:

• Payor rules - Managed by Silna based on each payor’s specific requirements for a given specialty and document type.
• Provider rules - Self-configured by providers in the Automation tab of the Provider Overview in the Silna portal. These let providers enforce their own documentation standards on top of payor requirements.

How It Works

Predictive Document Intelligence validates documents automatically when you upload them during a PATCH (Update) Prior Authorization request. Validation typically completes in 30 to 45 seconds.

Validation Flow

Validation failures do not prevent you from moving forward with creating a prior authorization. If you choose to proceed despite failures, you can either disable validation for specific document types or set skip_document_intelligence: true on your next PATCH request. The prior authorization will then advance to ALL_REQUIRED_DATA_RECEIVED as normal.

Integration Steps

1. Upload documents - Call PATCH Prior Authorization with clinical documents attached to the appropriate document types.
2. Poll for results - Call GET Prior Authorization with with_requirements=true. Once the status moves to DATA_REQUESTED or ALL_REQUIRED_DATA_RECEIVED, validation results are available in required_data.documents[].document_validation_batch.
3. Handle failures - If document_validation_batch.result is "FAILED", use the batch ID to get detailed rule results. You can fix documents and re-upload, or choose to proceed despite the failures (see Overriding Validations).
4. Complete the request - Once the status is ALL_REQUIRED_DATA_RECEIVED (either because all validations passed or because you chose to proceed), call POST Complete Prior Authorization.

Rule Types

Predictive Document Intelligence evaluates documents across seven rule categories. Both payor rules and provider-configured rules use the same rule types:

Rule Type	What It Validates	Example Checks
CLINICIAN	Clinician credentials and identity	Valid signature present, matching NPI/credential, correct taxonomy code
TIME_FRAME	Document dates and expiration	Document not expired, evaluation within required window, dates align with auth period
DIAGNOSIS	Diagnosis codes and dates	Required ICD codes present, diagnosis date documented
TREATMENT	Treatment and referral information	Referred specialties match requirements, CPT codes present
PATIENT	Patient demographics and history	Age verification, clinical history present, patient identity confirmed
ASSESSMENT	Diagnostic tools and clinical criteria	Required assessment instruments used, diagnostic criteria met, severity documented
DOCUMENT	Document type and completeness	Document matches the required type

Reading Validation Results

From the Prior Authorization

When you call GET Prior Authorization with with_requirements=true, each document in required_data.documents[] includes:

{
  "document_type": "DIAGNOSIS_REPORT",
  "required": true,
  "requirement_satisfied": false,
  "document_validation_batch": {
    "id": "0194a4ed-130d-775b-81fd-ca2b4eb618ce",
    "result": "FAILED"
  }
}

• document_validation_batch.result - "PASSED" or "FAILED"
• document_validation_batch.id - Use this ID to fetch detailed rule results via the Get Document Validation Batch endpoint
• requirement_satisfied - Set to false when validation fails. The prior authorization will stay in DATA_REQUESTED until the documents are fixed or validation is overridden

The document_validation field on the GET Prior Authorization response is deprecated. It only returns a single document validation and does not reflect the full batch. Use document_validation_batch instead, which provides the batch ID and aggregate result.

Detailed Rule Results

Use the Get Document Validation Batch endpoint to retrieve the full breakdown of which rules passed and failed:

{
  "id": "batch-uuid",
  "document_type": "DIAGNOSIS_REPORT",
  "result": "FAILED",
  "document_validations": [
    {
      "id": "validation-uuid",
      "result": "FAILED",
      "passed_rules": [
        {
          "rule": "DIAGNOSIS_CODES",
          "result": "PASSED",
          "reason": "Required ICD-10 codes F84.0 found in document"
        }
      ],
      "failed_rules": [
        {
          "rule": "DOCUMENT_EXPIRED",
          "result": "FAILED",
          "reason": "Document date (2024-01-15) exceeds the 90-day validity window"
        }
      ],
      "rule_results": [...]
    }
  ]
}

Each rule result includes a human-readable reason explaining why it passed or failed.

Use document_validation_batch.id from the GET Prior Authorization response to retrieve detailed results via the batch endpoints. The document_validation field on the PA response is deprecated — it only returns a single document validation. You can still use the individual document validation endpoint to look up a specific validation by ID.

Overriding Validations

Validation is designed to help catch issues before submission, but it never prevents you from creating a prior authorization. If you want to proceed despite failures, there are two options:

Option 1: Disable validation for specific document types

Set document_intelligence_configurations[].document_validation.enabled to false in your PATCH request to skip validation for individual document types while still validating others:

{
  "documents": [
    {
      "document_type": "REFERRAL",
      "file_id": "your-file-id"
    }
  ],
  "document_intelligence_configurations": [
    {
      "document_type": "REFERRAL",
      "document_validation": {
        "enabled": false
      }
    }
  ]
}

Option 2: Skip all validation

Set skip_document_intelligence to true in your PATCH request to bypass validation entirely for that update:

{
  "skip_document_intelligence": true,
  "documents": [
    {
      "document_type": "REFERRAL",
      "file_id": "your-file-id"
    }
  ]
}

In either case, the prior authorization will advance to ALL_REQUIRED_DATA_RECEIVED once all other requirements are met, regardless of any previous validation failures.

Using PDI Without Submitting a Prior Authorization

Predictive Document Intelligence runs as part of the prior authorization workflow, but you are not required to submit the authorization to Silna after validation completes. This is useful when you only want to validate documents — for example, to check clinical documentation quality before deciding whether to proceed.

1. Create a prior authorization — Call POST Create Prior Authorization as usual.
2. Upload documents — Call PATCH Update Prior Authorization with your documents.
3. Poll for validation results — Call GET Prior Authorization with with_requirements=true to retrieve the validation results from document_validation_batch.
4. At this point, you can either save the prior authorization as a draft, making it visible in the “Drafts” tab in the Silna Web App, or you can abandon the authorization if desired. To save as draft, you would use the PATCH API in Step 2 but only set the field save_as_draft to be true. If instead you want to abandon the prior authorization you should call DELETE Abort Prior Authorization to remove the authorization entirely. This is available while the PA is in the DATA_REQUESTED, DATA_RECEIVED, or ALL_REQUIRED_DATA_RECEIVED statuses.

Whether you are saving the prior authorization as a draft or abandoning it all together after running Predictive Document Intelligence, you should not call POST Complete

Prior Authorization Lifecycle

Validation runs automatically after you upload documents and before the authorization can be completed. When rules fail, the response includes details about which rules failed and why. You can fix and re-upload, or skip validation and proceed.

API Endpoints

Endpoint	Description
Get Document Validation Batch	Returns a batch with all document validations and rule results
Get Document Validations in a Batch	Returns paginated document validations within a batch
Get Document Validation	Returns a single document validation by ID