Features

Reader Vladimir Slamecka from Georgia Tech sent the following note about the history of "database mining".

In 1974, while chairing of the Georgia Tech computer science school, I wrote and submitted, jointly with the Emory Univ. School of Medicine, a research proposal to the National Library of Medicine. The proposal made a case for an alternate approach to computer-assisted diagnosis (alternate to the tree-based decision algorithms made popular by Stanford), based on deriving descriptions of patient analogs from a large aggregate database of empirical medical information (i.e., electronic patient records). In one of the support arguments for its development, I argued that given such a large database it would be possible to use it for "the generation of new knowledge" [sic], by automatically searching it for previously unknown data relationships and co-occurrences.

The proposal was funded for three years. Today, the resulting aggregate database of patient records is fully operational at the Grady Memorial Hospital in Atlanta. Arguably, it may be the largest aggregate EMR database in the world: over 58 million records emanating from encounters with 1.2 million patients are available online.

I was reminded of this when I saw your recent note about the history of the phrase "database mining". In our case, the concept and mention of knowledge discovery from data bases go back to around 1972.