Summary

Claim AI-based Adjudication System

...

The AI input model is represented by a matrix having on each row one Medical Item/Service from a Claim. If the FHIR Claim Bundle contains multiple Claims then the AI input model will contain all Items/Services from all Claims. Each row contains the following informations (either FHIR Medical OR ActivityDefinition, representing an Item in FHIR Claim):

...

Several data files are to be loaded and saved during the AI Model execution:

• encoded fields data file

• scalling file

• AI model file

The AI Model file name with full path (or relative path) is configured in the modules configuration key ai_model_file. Each time the model is executed, the model file is loaded, allowing to change the filename on execution in the configuration (hot-reload).

...

The following module configuration is defined:

{
    'authentication': [string], // Authentication key list. Default:String
 []   ai'claim_modelresponse_fileurl': string'http://localhost:8000/claim_ai/ClaimResponse', // AI Model file name with full path. Default: ./ai_model.pkl
  claim_response_organization: string // Value to build the insurer reference in the ClaimResponse. Default: openIMIS-Claim-AI
}URL set in the ClaimResponse. String
    'ai_model_file': "joblib_Voting_Model.pkl", // AI Model file name, relative location to module folder. String
    'ai_scaler_file': "scaler.pkl", // AI Scaler file name, relative location to module folder. String
    'ai_encoder_file': "Encoder.obj", // AI Encoder file name, relative location to module folder. String
    'claim_response_organization': 'openIMIS-Claim-AI', // Organisation name in ClaimResponse. String
    'date_format': '%Y-%m-%d', // Date format to use. String
    'first_date': '2016-01-01' // First date for date transformation. String
}

...

FHIR Contained Resources feature allows to integrate the referred resourced. Because AI-model is requiring information from other resources referenced by the Claim, these resources must to be accessible from the Claim sent by Claim-AI Quality module. In case of FHIR Claim, these are Patient, Condition, Medication, HealthcareService, Practitioner, ActivityDefinition. The mechanism to build Claim resource with Contained Resources is developed in the openIMIS FHIR R4 module. In FHIR R4 module, the inclusion of Contained Resources should be possible based on the query variable contained=true/false (default false).

...

@startuml

state "Enter Claim" as ClaimEntering {

 [*] --> Entered : Enter Claim

 
 Entered --> Entered : Edit\nClaim

 Entered : ClaimStatus=Entered (2)

 Entered : ReviewStatus=Idle (1)

 Entered : ClaimItemStatus=Accepted (1)

 Entered : RejectionReason=Accepted (0)

}

state "Rule Engine Checking" as RuleEngineCheck {

 state RuleEngineValidation <<choice>>

 Entered --> RuleEngineValidation : Submit Claim

 RuleEngineValidation --> RejectedByRuleEngine : Reject\nAll Items

 RejectedByRuleEngine : ClaimStatus=Rejected (1)

 RejectedByRuleEngine : ClaimItemStatus=Rejected (2)

 RejectedByRuleEngine : RejectionReason=[1-19]

 RejectedByRuleEngine : json_ext.claim_ai_quality.was_categorized=false

 RejectedByRuleEngine : json_ext.claim_ai_quality.

request_

time=ValidityFrom
 RejectedByRuleEngine : json_ext.claim_ai_quality.

response_

time=ValidityFrom
 RejectedByRuleEngine : json_ext.claim_ai_quality.ai_result=Rejected (2)

To ease the conversion from FHIR ClaimResponse to openIMIS Claim and saving into the DB, we have implemented the HTTP PUT method for ClaimResponse resource into openIMIS FHIR R4 Module.

Models

The new Claim-AI adjudication process is executed between Rule-based engine and the manual adjudication (see Getting the ClaimResponse from Claim-AI Module section). Because we decided not to add additional statuses to Claim, requiring to modify the Claim module, we will to use the JSON extensions that can be added and updated from the Claim-AI Quality module.

The following JSON extensions will be created to support the Claim AI-based Categorisation:

• to add information on Claim. The field was_categorized allow to filter automatically checked Claim that have not been categorized by AI and ‘hide’ them for manual adjudication.

  Code Block
  language json
  "json_ext": { "claim_ai_quality": { "was_categorized": "boolean", // whether the claim has been categorized by Claim-AI module or not "request_time": "datetime", // time when the claim has been sent to Claim-AI module "response_time": "datetime" // time when the claim adjudication response has been received from Claim-AI module } }

• to add information on ClaimItem and ClaimService. This is used to store AI categorisation result to allow the misclassification report.

  Code Block
  "json_ext": { "claim_ai_quality": { "ai_result": "integer" // claim adjudication value provided by Claim-AI module (for misclassification report) } }

...

 RejectedByRuleEngine --> [*]

 RuleEngineValidation --> CheckedIdle : Accept \n Some Items
 CheckedIdle : ClaimStatus=Checked (4)
 CheckedIdle : json_ext.claim_ai_quality.was_categorized=false
 
}

state "AI Automatic Categorisation" as AICheck {

 state AIClaimCategorisation <<choice>>

 CheckedIdle --> AIClaimCategorisation 

 AIClaimCategorisation --> CheckedAIAccepted : Accept\nAll Items
 CheckedAIAccepted : json_ext.claim_ai_quality.was_categorized=true
 CheckedAIAccepted : json_ext.claim_ai_quality.ai_result=Accepted (1)

 AIClaimCategorisation --> CheckedAIFlagged : Reject\nSome Items
 CheckedAIFlagged : ReviewStatus=SelectedForReview (4)
 CheckedAIFlagged : ClaimItemStatus=Rejected (2)
 CheckedAIFlagged : RejectionReason=Rejected by AI (-2)
 CheckedAIFlagged : json_ext.claim_ai_quality.was_categorized=true
 CheckedAIFlagged : json_ext.claim_ai_quality.ai_result=Rejected (2)

 note top of CheckedAIFlagged : Rejected Items values 

}

state "Manual Review" as ManualReview {

 state SelectForReview <<choice>>

 CheckedAIAccepted --> SelectForReview

 SelectForReview --> CheckedNotSelected : Select manually\nnot to review
 CheckedNotSelected : ReviewStatus=Not Selected (2)
 CheckedNotSelected : RejectionReason=Accepted (0)

 SelectForReview --> CheckedSelected : Select manually\nto review
 CheckedSelected : ReviewStatus=Selected (4)
 CheckedSelected : RejectionReason=Accepted (0)

 state RequireManualReview <<choice>>

 CheckedSelected --> RequireManualReview 

 CheckedAIFlagged --> RequireManualReview : Selected automatically\nfor review

 RequireManualReview --> CheckedRejected : Manual Review\nReject All Items
 CheckedRejected: ClaimStatus=Rejected (1)
 CheckedRejected: ClaimItemStatus=Rejected (2)
 CheckedRejected : RejectionReason=Rejected by MO (-1)

 CheckedRejected --> [*]

 RequireManualReview --> CheckedAccepted : Manual Review\nAccept Some Items
 CheckedAccepted : ClaimItemStatus=Accepted (1)
 CheckedAccepted : RejectionReason=Accepted (0)

 CheckedAccepted  --> CheckedReviewed : Deliver review
 CheckedReviewed : ReviewStatus=Reviewed (8)

}

state "Processing" as Process {

 CheckedReviewed --> ValuatedAccepted : Process
 ValuatedAccepted : ClaimStatus=Valuated (16)

 CheckedNotSelected --> ValuatedAccepted : Process

 CheckedAccepted --> ValuatedBypassed : Process \nwithout \ndelivering \nreview
 ValuatedBypassed : ClaimStatus=Valuated (16)
 ValuatedBypassed : ReviewStatus=Bypassed (16)

 ValuatedBypassed --> [*]

 ValuatedAccepted --> [*]

}

@enduml

To ease the conversion from FHIR ClaimResponse to openIMIS Claim and saving into the DB, we have implemented the HTTP PUT method for ClaimResponse resource into openIMIS FHIR R4 Module.

Models

The new Claim-AI adjudication process is executed between Rule-based engine and the manual adjudication (see Getting the ClaimResponse from Claim-AI Module section). Because we decided not to add additional statuses to Claim, requiring to modify the Claim module, we will to use the JSON extensions that can be added and updated from the Claim-AI Quality module.

The following JSON extensions will be created to support the Claim AI-based Categorisation:

• to add information on Claim. The field was_categorized allow to filter automatically checked Claim that have not been categorized by AI and ‘hide’ them for manual adjudication.

  Code Block
  language json
  "json_ext": { "claim_ai_quality": { "was_categorized": "boolean", // whether the claim has been categorized by Claim-AI module or not "request_time": "datetime", // time when the claim has been sent to Claim-AI module "response_time": "datetime" // time when the claim adjudication response has been received from Claim-AI module } }

• to add information on ClaimItem and ClaimService. This is used to store AI categorisation result to allow the misclassification report.

  Code Block
  "json_ext": { "claim_ai_quality": { "ai_result": "integer" // claim adjudication value provided by Claim-AI module (for misclassification report) } }

Because currently ClaimItem and ClaimService doesn’t support custom fields, these classes will also extend the ExtendableModel class.

Migrations

Because the Claim-AI modules are activated after Claims have been adjudicated (the data used for the training), a migration script needs to be created to insert the JSON fields into the available data. The following initializations will be used:

If Claim.ClaimStatus == Entered (Value 2) Then don’t add Claim.json_ext.claim_ai_quality

If Claim.ClaimStatus == Checked (Value 4) Then add

Claim.json_ext.claim_ai_quality.

was_categorized = false
Claim.json_ext.claim_ai_quality

.request_time = null
Claim.json_ext.claim_ai_quality.

Because currently ClaimItem and ClaimService doesn’t support custom fields, these classes will also extend the ExtendableModel class.

Migrations

...

response_time = null

If Claim.ClaimStatus == Rejected (Value 1) Then add

Claim.json_ext.claim_ai_quality.was_categorized = false
Claim.json_ext.claim_ai_quality.request_time = Claim.ValidityFrom
Claim.json_ext.claim_ai_quality.response_time = Claim.ValidityFrom
ClaimService.json_ext.claim_ai_quality.ai_result = ClaimService.ClaimServiceStatus
ClaimItem.json_ext.claim_ai_quality.ai_result = ClaimItem.ClaimItemStatus

If Claim.ClaimStatus in [Reviewed/Processed (Value 8), Valuated (Value 16)] Then add

Claim.json_ext.claim_ai_quality.was_categorized = true
Claim.json_ext.claim_ai_quality.request_time = Claim.ValidityFromReview
Claim.json_ext.claim_ai_quality.response_time = Claim.ValidityFromReview
ClaimService.json_ext.claim_ai_quality.ai_result = ClaimService.ClaimServiceStatus
ClaimItem.json_ext.claim_ai_quality.ai_result = ClaimItem.ClaimItemStatus

...

• ClaimFilterMedicalItem allows Medical Officers to filter the Claims list based on a specific Medical Item (only one from autocomplete list)

• ClaimFilterMedicalService allows Medical Officers to filter the Claims list based on a specific Medical Service (only one from autocomplete list)

• ClaimFilterAIProcessed allows Medical Officers to display or hide the non AI-processed Claims (checkbox, default checked)dropdown with options: Show all, AI-Categorized Claim Only, Not AI-Categorized Claims Only, default AI-Categorized Claim Only)

Custom Claim Review Search Results - Frontend

The following column will be added to the claim.Searcher result list:

• ClaimResultsAIProcessed will display the value of Claim.json_ext.claim_ai_quality.was_categorized as True/False.

AI Categorisation Misclassification Report

...

• total number of Claims

• accuracy scoresscore ( Accuracy = (TP+TN)/(TP+TN+FP+FN) )

• number of True Positives results

• number of True Negative results

• number of False Positive results

• number of False Negative results

• Table with Claim that were fixed (ClaimItem.json_ext.claim_ai_quality.ai_result != ClaimItem.ClaimItemStatus - same for ClaimService)

  • same format as Claim Overview report with additional Scope as ‘Claim and All Details’

  • For each Item and Service will be added:

    • json_ext.claim_ai_quality.ai

    _result column for each Item and Service

    • _result column

    • Medical Officer name from AuditUserIDReview (same as for ClaimAdmin)

To generate this report, the user will have have to filter through the Review Claims search form and trigger the report through a button (label is “AI Misclassification Report”). Only the filtered Claims will be considered for the report.

...

@startuml

node openIMISImplementation as "openIMIS Implementation" {

component ClaimModule as "Claim Module"

component ClaimAIQualityModule as "Claim AI Quality Module"

component FHIRR4Module as "FHIR R4 Module"

FHIRR4Module -[dashed]-> ClaimModule: <<use>>

ClaimAIQualityModule -[dashed]-> FHIRR4Module : <<use>>

ClaimAIQualityModule --^ ClaimModule

database openIMISDB

ClaimModule -[dashed]-> openIMISDB:

 <<use>>

}


node openIMISClaimAIImplementation as "openIMIS Claim-AI Implementation" {

component ClaimAIModule as "Claim-AI Module"

artifact ClaimAIModel as "Claim-AI Model"

ClaimAIModule -[dashed]-> ClaimAIModel: <<use>>

}


ClaimAIModule -0)- ClaimAIQualityModule: FHIR Claim

ClaimAIModule -(0- ClaimAIQualityModule: FHIR ClaimResponse

node ClaimAITraining as "openIMIS Claim-AI Training" {

component ClaimAIAlgorithm as "Claim-AI ML Algorithm"

artifact openIMISDBReplica as "openIMIS Claim Data"

ClaimAIAlgorithm -[dashed]-> ClaimAIModel:

 <<generates>>
openIMISDBReplica -[dashed]-> openIMISDB: <<replicates>>
ClaimAIAlgorithm -[dashed]->

 openIMISDBReplica: <<use>>

}

@enduml

Testing

Testing the WebSocket from JavaScript console

...