Participants
- MartinPlayZ (I) - Interested in learning about medical AI and open-source solutions
- meephasquestions (meep) - Researcher passionate about learning tools and medical AI applications
Main Topics Discussed
1. WikiLinks App Discussion
What meep shared:
- WikiLinks was an iOS-only app that created visual mind maps while browsing Wikipedia
- Features that made it special:
- Automatic node creation for hyperlinked topics
- Visual web/mind map generation in real-time
- Automatic clustering and organization of related topics
- Ability to toggle between Wikipedia articles and visual overview
- Integrated photo and video search for topics
Why it was important:
- Perfect for neurodivergent learning styles
- Solved the “thousands of browser tabs” problem
- Automatically organized research connections
- Made complex topic relationships visible and navigable
Current status: App is no longer available, exported maps can’t be opened
2. Medical AI Research Discovery
Every Cure & MATRIX Platform:
- Founded by David Fajgenbaum (2022) after his personal experience with Castleman disease
- AI platform that matches FDA-approved drugs with diseases for repurposing
- Uses knowledge graphs and machine learning to calculate “predictive efficacy scores”
- Secured $48 million ARPA-H funding in 2024
- Public-facing tool planned for 2026
Current Limitations:
- MATRIX not yet publicly accessible
- Doesn’t currently model comorbidities or disabilities
- Focus is on drug-disease matching, not holistic patient context
3. Existing Medical AI Alternatives
Available Tools Mentioned:
- Clinical Decision Support Systems: Epic AI, IBM Watson Health, MedPaLM-M
- Drug Repurposing Platforms: BenevolentAI, Healx, Insilico Medicine
- Precision Medicine: Tempus, nference
- Research Tool: Consensus.app for searching scientific literature
4. Local Medical AI Setup Options
Downloadable Models (GGUF format):
- BioMistral (biomedical Q&A)
- MedAlpaca (clinical reasoning)
- Clinical LLaMA (medical diagnosis conversations)
- PMC-LLaMA (biomedical research papers)
Suggested Workflow:
- Use GPT4All or similar app with medical GGUF models
- Manually search PubMed for relevant abstracts
- Create structured prompts combining research abstracts with specific questions
- Get AI-assisted analysis of drug repurposing possibilities
Key Insights
Where We Are Now
- No single public tool combines drug repurposing + comorbidity modeling + disability considerations
- Proprietary medical AI exists but isn’t publicly accessible
- Open-source medical models are available but require manual integration with research databases
- Every Cure’s MATRIX represents the most promising future solution (2026 public release)
Immediate Actions:
- Try Consensus.app for medical research searches (meep’s recommendation)
- Set up local medical AI:
- Explore existing clinical tools if you have medical professional access
Medium-term Opportunities:
- Monitor Every Cure’s 2026 public release for comprehensive drug repurposing
- Experiment with RAG workflows combining local models + medical databases
- Consider developing a WikiLinks-style tool for medical research visualization
Long-term Vision:
- Combine visual mind mapping (like WikiLinks) with medical AI capabilities
- Create tools that consider whole-patient context (comorbidities, disabilities, quality of life)
- Contribute to open-source medical AI development
Personal Context
- I prefer open-source, self-hosted solutions
- I have programming knowledge but limited Python experience
- I’m interested in tools that organize complex information visually -> Obsidian
Next Steps for Local Experimentation
- Download and test the recommended medical GGUF models
- Practice creating effective prompts for medical research questions
- Experiment with PubMed + AI model workflows
- Consider building tools that bridge the gap between current capabilities and ideal functionality
Conversation Date: August 23, 2025