The problem

The digital health ecosystem is increasingly saturated with misleading and unverified medical information. With the rise of generative AI and short-form viral content, health-related claims spread faster than they can be validated.

Critical signals exist across multiple domains:

• Social media platforms (TikTok, X, Instagram) where claims originate and go viral.
• Scientific literature (peer-reviewed journals, PubMed).
• Public health institutions (CDC, WHO, NIH).
• Health guidelines and consensus reports.

However, these sources remain disconnected, making it extremely difficult to:

• Verify medical claims in real time.
• Distinguish between credible and misleading health information.
• Provide transparent, evidence-backed responses at scale.
• Track how misinformation evolves across platforms.

As a result:

• Harmful health narratives spread unchecked.
• Platforms lack scalable verification mechanisms.
• Users make decisions based on incomplete or false information.
• Trust in digital health information continues to decline.

The project goals

This project proposes building VeriHealth AI, a real-time medical misinformation detection system designed to connect viral health claims with validated scientific consensus.

The solution focuses on constructing a structured claim-to-evidence dataset and enabling AI-powered verification through Retrieval-Augmented Generation (RAG).

Key components include:

• Collecting viral medical claims from social media platforms.
• Aggregating scientific evidence from trusted public sources (PubMed, CDC, WHO, NIH).
• Designing a structured claim-to-evidence mapping framework.
• Developing a RAG-based fact-verification engine.
• Building a misinformation detection API for real-time use.
• Defining verification standards and traceability guidelines.

As part of this challenge, the system must demonstrate the ability to:

• Link unstructured social media claims to authoritative scientific evidence.
• Align multiple sources into a consistent verification framework.
• Classify claims based on accuracy, ambiguity, and risk level.
• Retrieve relevant, high-quality supporting evidence.
• Provide transparent traceability between claims and sources.
• Handle noisy, ambiguous, or partially true claims.
• Deliver real-time responses through an API interface.

Impact of the Problem

Digital Platforms & Social Media

• Scalable detection of harmful medical misinformation.
• Improved content moderation support with explainable AI.
• Reduced the spread of misleading health narratives.

Public Health Organizations

• Faster identification of emerging misinformation trends.
• Data-driven communication strategies.
• Stronger ability to respond to public health risks.

Researchers & AI Developers

• Access to high-quality, structured verification datasets.
• Foundation for building trustworthy AI systems.
• Acceleration of research in AI safety and fact-checking.

General Public

• Access to reliable, evidence-based health information.
• Increased trust in digital content.
• Reduced exposure to harmful or misleading advice.

Real-World Impact

• Reduction in the spread of harmful medical misinformation.
• Improved public health awareness and decision-making.
• Stronger alignment between digital content and scientific consensus.
• Advancement of transparent and trustworthy AI systems.

Timeline

1

Sprint 1: Data Discovery & Collection Setup (Weeks 1–2)

• Establishing the data acquisition pipelines for viral medical claims and trusted scientific sources.

2

Sprint 2: Dataset Structuring & Evidence Mapping (Weeks 3–4)
Building the structured dataset and defining workflows to map claims to validated evidence.

3

Sprint 3: Verification Intelligence Layer (Weeks 5–6)

• Developing the RAG-based verification engine and claim classification mechanisms.

4

Sprint 4: API Development & Final Delivery (Weeks 7–8)

• Delivering the real-time misinformation detection API, validating system performance, and finalizing documentation.

**More details will be shared with the designated team.