Lexically-based Distinction of Readability Levels of Health Documents

A method for automatically estimating the readability of medical websites has been designed and tested. Web Health documents address both medical professionals and non expert users. Their technical levels are heterogeneous. Such situations can hinder the understanding of documents by users who are new to a health topic. They may thus have difficulties in assimilating the content and furthermore could misinterpret its meaning. The first step for helping citizens becomes to let them choose those documents which are more suitable to their current knowledge on a given disease or health topic. The method is based on the analysis of the lexical clues and computing of how common and familiar is the vocabulary used. We address this issue within the Health on the Net database Foundation. Evaluation shows currently 92% accuracy.