[ADE-66] Enhance medical document processing by adding trusted medical sources for analysis and other improvements

[adamrefaey]

Change

This pull request introduces significant enhancements to the medical document processing pipeline by incorporating verification against trusted medical sources and improving error handling. Key changes include adding a verification step using Perplexity AI, updating the analysis logic to handle missing reference ranges, and expanding the Perplexity service with a new method for reviewing medical analyses.

Enhancements to Medical Document Analysis:

• Updated aws-bedrock.service.ts to include a list of trusted medical sources (e.g., Mayo Clinic, CDC, WHO) for analysis and to handle cases where reference ranges are missing by using standard ranges from these sources. ([[1]](https://github.com/ModusCreateOrg/app-med-ai-gen/pull/110/files#diff-5c85186b48ca73f6f71ee9bca3b21941f8a85b0f18c47f20c0bb94dfddcbb0adR59-R70), [[2]](https://github.com/ModusCreateOrg/app-med-ai-gen/pull/110/files#diff-5c85186b48ca73f6f71ee9bca3b21941f8a85b0f18c47f20c0bb94dfddcbb0adR103))

Integration of Perplexity AI for Verification:

• Modified document-processor.service.ts to introduce a new step where Perplexity AI reviews and verifies the medical analysis. If the review fails, the system falls back to the initial analysis. ([backend/src/document-processor/services/document-processor.service.tsL58-R82](https://github.com/ModusCreateOrg/app-med-ai-gen/pull/110/files#diff-5e997b182ce449d133c60c45abf8b6a3951b5730d733d3cc56869f4d5624df89L58-R82))
• Added a new reviewMedicalAnalysis method in perplexity.service.ts to validate medical document analyses against trusted sources, ensuring accuracy in lab values, interpretations, and recommendations. ([backend/src/services/perplexity.service.tsR165-R229](https://github.com/ModusCreateOrg/app-med-ai-gen/pull/110/files#diff-43cc86a138bda7d361d75e44ee9b0ff5c3128c44e30d9f7bc4cbc01a8ac6b45dR165-R229))

Improvements to Perplexity API Interaction:

• Simplified the createChatCompletion method's description in perplexity.service.ts to clarify its purpose. ([backend/src/services/perplexity.service.tsL99-R99](https://github.com/ModusCreateOrg/app-med-ai-gen/pull/110/files#diff-43cc86a138bda7d361d75e44ee9b0ff5c3128c44e30d9f7bc4cbc01a8ac6b45dL99-R99))

Does this PR introduce a breaking change?

{...}

What needs to be documented once your changes are merged?

{...}

Additional Comments

{...}

[Copilot]

Pull Request Overview

This PR enhances the medical document processing pipeline by integrating trusted medical sources verification and improving error handling with Perplexity AI.

• Incorporates a review step using Perplexity AI in the document processing flow.
• Updates AWS Bedrock analysis instructions to include a comprehensive list of trusted medical sources and handle missing reference ranges.
• Introduces a new reviewMedicalAnalysis method in the Perplexity service for verifying and correcting medical analyses.

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
backend/src/services/perplexity.service.ts	Added reviewMedicalAnalysis method and updated createChatCompletion documentation.
backend/src/document-processor/services/document-processor.service.ts	Integrated Perplexity review step with error fallback.
backend/src/document-processor/services/aws-bedrock.service.ts	Updated extraction instructions to incorporate additional trusted medical sources and reference range handling.