Refine relevance selection in MedicalAdviceGenerator to clarify intent in chunk selection

Merge branch 'newbranchYB-newest' into Merged20250805

Add anaphylaxis condition mapping to keyword configuration

Merge branch 'newbranchYB-newest' into Merged20250805

Add adaptive relevance thresholds for query complexity in PrecisionMRRAnalyzer; fix typo in condition mapping for postpartum hemorrhage

Refactor relevance calculation and update thresholds in latency evaluator; enhance precision and MRR analyzer with angular distance metrics; increase timeout for primary generation in fallback configuration.

Add multi-system evaluation support for clinical actionability and evidence quality metrics

Fix medical condition extraction and validation logic across multiple levels

Fix JSON parsing error in medical condition extraction

feat(app): add multi-mode retrieval system with critical bug fixes

fix(mild bug): enhance user query prompts (more robust dealing process with .txt or .json) and add postpartum hemorrhage condition mapping

feat: Simplify fallback architecture and remove length restrictions

Merge branch 'main' into Jeff

fix(MedicalAdviceGenerator): enhance fallback generation with primary result simplification and increased token limits

fix(MedicalAdviceGenerator): update fallback token limits and response guidelines for improved clarity and guidance

fix(llm-clients): relax abnormal response detection threshold for medical advice

fix(customization): implement anti-hallucination filtering with Top-P sampling and similarity thresholds

Merge branch 'main' into Jeff

feat: Implement 3-tier fallback generation mechanism in MedicalAdviceGenerator

Pull newest code from main

refactor: enhance example queries and prompt structure for improved medical guidance clarity

feat: implement flexible condition extraction and regex matching in user queries, to make fallback userprompt more robust

feat(integration): integrate hospital customization pipeline with main RAG system

feat: enhance medical query processing with user-friendly output and debug mode handling

feat and fix: update medical advice extraction logic to prioritize raw response

Add end-to-end pipeline test script for OnCall.ai

refactor: improve multilayer logic in llm_clients to to enhance functionality and improve performance based on medical or unmedical issues

Re-check multilevel fallback validation test suite for OnCall.ai

feat(user_prompt): Enhance Medical Query Processing Pipeline

feat(user_prompt): update UserPromptProcessor to integrate Llama3-Med42-70B and enhance query validation; add unit tests for condition extraction and matching mechanisms

feat(retrieval): add sliding window search method for enhanced semantic search

feat(user_prompt): add keyword index checking and enhance medical query validation (add validate_medical_query() and _check_keyword_in_index(), based on implementation_todo_20250730_user_prompt.md

feat(llm_clients): enhance MeditronClient to support local model loading and improve error handling

WIP(llm_clients): add MeditronClient for medical query processing with Hugging Face integration

🚀 Implement Advanced Condition Extraction for Medical Query Processing

refactor(deduplication): change deduplication logic from distance-based to exact text matching

feat(retrieval) 1st MVP: enhance search logging and deduplication logic with distance threshold

WIP: feat(retrieval): implement basic vector retrieval system for medical documents

refactor(data_processing): enhance chunking and embedding generation

WIP: Enhance dual keyword chunking to include pre-calculated metadata for treatment chunks

feat(data_processing): Implement token length control with semantic preservation

refactor(data_processing): add token-based chunking strategy for improved keyword context

refactor(data_processing): optimize chunking strategy with token-based approach

feat(data-processing): implement data processing pipeline with embeddings