The SQLX system: generating explanations for clinical rules encoded in SQL.

Abstract We have developed a system to provide case-specific explanations for SQL query results. The explanation facility, called SQLX, is intended to improve user acceptance of clinical rules that are encoded in the SQL query language and are triggered by updates to a relational database. Our design goal is to combine the efficiency and familiarity of the SQL query model with a more intuitive and case-specific display of query results. SQLX defines an augmented SQL syntax that allows the incorporation of explanation text directly into query specifications. A query analyzer determines which retrieved data logically contribute to rule firing, and combines the data with explanation text to generate a case-specific explanation. Although features of the SQL query model limit the use of this method as a general-purpose rule-explanation facility, the current design accommodates a wide range of SQL query formulations and can provide efficient processing and intuitive explanations of many clinical rules.