Specialized tools are needed when searching the web for rare disease diagnoses

Area of Science:

• Medical Informatics
• Information Retrieval
• Rare Diseases

Background:

• Diagnosing rare diseases presents significant challenges.
• Web search engines like Google Search and PubMed are often used but may lack specialized capabilities.

Purpose of the Study:

• To evaluate the performance of Google Search and PubMed in rare disease diagnosis.
• To introduce and assess FindZebra, a specialized search engine for rare diseases.

Main Methods:

• Developed an evaluation framework with 56 diagnostic cases.
• Created FindZebra, a specialized search engine using curated rare disease sources and the Indri information retrieval system.
• Compared the diagnostic performance of FindZebra against Google Search and PubMed.

Main Results:

• FindZebra demonstrated superior performance compared to Google Search and PubMed in diagnosing rare diseases.
• Evaluation results and subsequent feedback confirmed FindZebra's effectiveness.

Conclusions:

• Specialized search engines like FindZebra offer advantages over general search engines for complex diagnostic tasks.
• General web search is not optimized for the specific needs of rare disease diagnosis.
• Emerging trends involve leveraging web resources and social media for medical diagnosis.