FindZebra: a search engine for rare diseases

Area of Science:

• Medical Informatics
• Bioinformatics
• Computational Biology

Background:

• The internet is a key resource for health information, with web search as the primary access method.
• Diagnosing rare diseases presents challenges due to symptom diversity and scattered information.
• Evaluating web search engine performance for rare disease diagnostics is crucial.

Purpose of the Study:

• To evaluate web search engines for rare disease diagnosis.
• To introduce and assess FindZebra, a specialized search engine for rare diseases.
• To develop an evaluation framework for rare disease diagnostic search tools.

Main Methods:

• An evaluation approach using 56 real diagnostic cases was designed.
• Google Search was customized, and a specialized rare disease search engine, FindZebra, was developed.
• FindZebra utilizes open-source technology and curated online medical information.

Main Results:

• FindZebra outperformed Google Search in diagnostic accuracy, both in default and customized settings.
• Specialized functionalities were added to FindZebra, leveraging medical ontologies and UMLS concepts.
• Methods for displaying retrieved results to medical experts were demonstrated.

Conclusions:

• Specialized search engines can enhance diagnostic quality for rare diseases.
• FindZebra offers improved diagnostic capabilities without compromising ease of use.
• The evaluation approach provides a benchmark for future search engine development.