Consortium.Doctor

• 

  logotype

Inspiration

The inspiration for Consortium.Doctor is a personal mission to prevent medical tragedies. After losing my mother to an undetected early-stage cancer and personally enduring a five-year misdiagnosis of "neck issues" that were actually migraines, I realized the human cost of diagnostic oversights. Building this system in Western Ukraine—navigating electricity cutoffs and the urgent needs of hospitals during wartime—has added a layer of profound urgency.

What it does

Consortium.Doctor acts as an AI-powered medical safety net. It provides a "consortium" of AI-driven perspectives to review symptoms and medical data, specifically looking for early-stage red flags that a single practitioner might overlook. The platform features:

• Internationalization: Support for 7 languages to ensure accessibility for diverse populations.

• AI-Generated Presentations: Digital doctor avatars that narrate findings through empathetic video explanations.

• File Export: Allows users to export their resulting diagnosis analysis as a document to share with their primary care physicians or specialists.

How we built it

The platform is a high-performance implementation of the Gemini ecosystem, designed to be resilient and scalable:

• Backend Intelligence: Gemini serves as the primary reasoning engine, tasked with deep medical cross-referencing and logic.

Development Flow: I utilized Claude Code for rapid, iterative coding to maximize productivity during limited power windows.

• Frontend & Deployment: The system is built for global access and deployed on Vercel.

Multimedia Integration: I used AI video tools to offer a "doctor narrative" in video format, presenting the project's findings in a way that feels human and accessible.

Challenges we ran into

The primary challenge was ensuring Diagnostic Integrity and overcoming "Model Bias"—the tendency for AI to agree with a provided premise. To solve this, I used ChatGPT as an adversarial agent to generate "confidently wrong" diagnosis lists. I then tested Gemini to see if it would catch these errors. Engineering the system to remain skeptical and prioritize evidence over hallucinations was a significant technical hurdle, made even more difficult by the infrastructure instability and power outages in Ukraine.

Accomplishments

Successfully launching consortium.doctor under wartime conditions and proving that the system can debunk adversarial AI-generated decoys are milestones that validate the project's potential to save lives.

What we learned

This project taught me that the best medical AI must be a "skeptical specialist." We learned that the "missing link" in healthcare isn't always a lack of data, but a lack of a comprehensive review process that questions initial assumptions. Furthermore, building during a conflict reinforced that AI can be a critical tool for resilience, providing expert-level second opinions when local resources are stretched thin.

What's next for Consortium.Doctor

The roadmap is focused on clinical trust and global expansion:

• Multi-Model Consensus: Integrating additional LLMs and custom models to work alongside Gemini for a true "Consortium" of verification.

• Clinical Integration & API: Passing official medical tests to offer an API that clinics in Ukraine, the US, and elsewhere can integrate into their own systems.

• Medical Accuracy: Adapting a list of actual, proven medications and diagnostic tests to ensure all narratives are grounded in the latest clinical standards.

• Global Availability: Scaling the marketing and infrastructure so this safety net is available to everyone, everywhere.

Built With

• claudecode
• gemini
• github
• nextjs
• node.js
• react
• vercel