MediSearch: Advanced Medical Web Search Engine

Generic web search engines like Google and Bing are helpful for daily tasks and finding answers to most general questions. However, they could perform better for domain-specific queries. For the medical field, MayoClinic provides easy to use interface for common diseases, FindZebra is excellent for rare diseases search, and MediCrawl performs well for both rare and common diseases. In this paper, we propose MediSearch to extend the capabilities of the MediCrawl web search engine for diseases with image classification and textual queries. We propose to use machine learning algorithms on skin pictures taken from digital cameras or bones and tissue pictures taken from radiology devices, along with the free text search capabilities.

I. Introduction

The internet (world wide web) has become a primary information source for our day-to-day tasks, including illnesses and treatments for medical professionals. Web search engines are the dominant interfaces to this information. While popular web search engines like Google, Bing, Yahoo, DuckDuckGo, and Ecosia [1]–[3] work well for most general queries, they are not designed to be superior in a particular domain. Hence they are generic. Unsurprisingly, they need to work better for diagnostic queries for medical diseases. A specialized search engine can improve diagnostic quality without devaluing the ease of use of standard web search engines. Niche tools like MayoClinic [4], FindZebra [5], and MediCrawl [6] aim to solve this problem by catering to medical diagnosis by providing domain-specific results.