Incorporating Bigram Statistics to Spelling Correction Tools

The study presented here describes an algorithm designed to automatically determine the correct spelling of misspelled words found in clinical notes. Based on the list of suggested spelling correction from a spelling suggestion tool, the correct suggestion is determined by incorporating the content words surrounding the misspelled word. The content words and the suggested spelling corrections form bigrams in which measures of association can be performed to determine the likelihood of a suggested word being the correct spelling.