BMLibrarian

The Biomedical Researcher's AI Workbench

BMLibrarian is a comprehensive AI-powered platform designed to be a complete workbench for biomedical researchers, clinicians, and systematic reviewers. It provides evidence-based answers to clinical questions, peer-review quality automated assessment of research papers, and systematic fact-checking of biomedical statements—all powered by local AI models requiring no cloud APIs or external services.

Related Projects

BMLibrarian is part of a family of open-source tools for biomedical research. These companion projects extend its capabilities in different directions:

• bmlibrarian_lite — A lightweight version of BMLibrarian that runs without PostgreSQL. Features native desktop (PySide6), iOS/macOS (Swift), and Android (Kotlin) apps with SQLite+sqlite-vec storage, FastEmbed embeddings, and support for multiple LLM providers (Anthropic, OpenAI, Ollama).

• bmlib — Shared Python library providing LLM abstraction across providers (Anthropic, OpenAI, Ollama, DeepSeek, Mistral, Gemini), quality assessment pipelines, transparency analysis, publication ingestion from PubMed/bioRxiv/medRxiv, and database utilities. Available on PyPI.

• biasbuster — Toolkit for building curated training datasets to fine-tune LLMs for detecting bias in biomedical abstracts. Includes data collection from Retraction Watch and Cochrane, multi-dimensional bias taxonomy, LoRA fine-tuning infrastructure, and evaluation harness with statistical comparison of model performance.

• BioMedicalNews — Biomedical news reader that fetches preprints from medRxiv, bioRxiv, and Europe PMC, uses LLM-based relevance scoring and quality evaluation, and delivers curated digests via email. Built on bmlib.

Why BMLibrarian?

Evidence-Based Answers to Clinical Questions

Ask questions like "What are the cardiovascular benefits of exercise?" or "Does metformin reduce mortality in diabetic patients?" and receive comprehensive, citation-backed reports synthesizing evidence from the latest biomedical literature.

Automated Research Quality Assessment

Evaluate research papers with the rigor of peer review:

• Paper Weight Assessment: Evaluate the evidential weight of studies based on study design, sample size, methodological quality, and risk of bias
• PRISMA 2020 Compliance: Assess systematic reviews against the 27-item PRISMA 2020 checklist
• PICO Extraction: Automatically extract Population, Intervention, Comparison, and Outcome components for systematic reviews

Robust Fact-Checking

Validate biomedical statements with literature evidence:

• Statement Fact-Checker: Evaluate claims like "Vaccine X causes Y" against published literature
• PaperChecker: Validate research abstract claims by systematically searching for contradictory evidence
• Counterfactual Analysis: Actively search for evidence that contradicts initial findings for balanced conclusions

Works Offline—Critical for Global Health

BMLibrarian is designed for clinicians and researchers working in areas with limited or unreliable internet connectivity:

• Runs entirely with local AI models via Ollama—no cloud APIs required
• Local database of PubMed and medRxiv publications with full-text PDFs where available
• No API keys, subscriptions, or external services needed
• Periodic synchronization with PubMed and medRxiv when connected
• Complete functionality offline after initial data import

This makes BMLibrarian uniquely valuable for healthcare workers in remote regions, field hospitals, developing nations, or any environment where reliable internet cannot be guaranteed.

Multiple Search Strategies with AI Assistance

BMLibrarian employs sophisticated multi-strategy search capabilities:

• Multi-model query generation: Use multiple AI models to generate diverse database queries (not currently wired into any app — see ROADMAP §1)
• Semantic search: Vector-based similarity search using document embeddings
• HyDE (Hypothetical Document Embeddings): Generate hypothetical answers to improve search relevance
• Keyword extraction: Traditional keyword-based search as fallback
• Counterfactual search: Actively search for contradictory evidence

Privacy-Preserving AI

All AI processing happens locally on your hardware:

• No data leaves your machine—perfect for sensitive patient data or pre-publication research
• No usage tracking or telemetry
• Complete control over model selection and parameters

What's New

Latest Features (02/2026):

Paper Reviewer Lab

A comprehensive paper assessment tool that combines all of BMLibrarian's analysis agents into a single unified workflow. Accepts input via DOI, PMID, PDF file, or pasted text.

# Launch the Paper Reviewer Lab
uv run python scripts/paper_reviewer_lab.py

11-Step Assessment Workflow:

1. Resolve Input (fetch document metadata via DOI/PMID/PDF/text)
2. Generate Summary (2-3 sentence synopsis)
3. Extract Hypothesis (identify core claims)
4. Detect Study Type (classify research methodology)
5. PICO Analysis (Population/Intervention/Comparison/Outcome)
6. PRISMA 2020 Assessment (systematic review checklist, if applicable)
7. Paper Weight Assessment (evidential weight scoring)
8. Study Quality Assessment (trustworthiness evaluation)
9. Synthesize Strengths/Weaknesses
10. Search Contradictory Evidence (optional PubMed search)
11. Compile Comprehensive Report

Key Features:

• Multiple input methods: DOI, PMID, PDF file, pasted abstract text
• Real-time workflow progress visualization (PySide6/Qt)
• Model selection from available Ollama models
• Results display in Markdown and JSON
• Export to Markdown, PDF, or JSON

This is the beginning of the README. Read the rest at the original. Read the full README ↗