Classification program Xmed

Based on powerful text analysis

Xmed of Gregori and Luz (doctoral thesis of Luz: 'Xmed vom Freitext zum Kode') is successor of the proven KLASS system for BAIK (see BAIK-Buch p 45). Xmed‘s special strength is free text classification. It includes a real translation program, TRANSSOFT of W. Moore, which first translates narrative into standardized German.

TRANSSOFT was developed by Moore for the translation of a German pathology textbook into English. Together Moore and I developed it for English to German and I used it for the successful translation of DXplain by Octo Barnett. Afterwards I used it for the German to German translation of narrative into standardized medical texts.

The translation included isolation of medical facts and their causal, temporal and intentional relations. These medical facts and their relations served as input to the rule based classification branches for Diagnoses (ICD-10) and procedures (IKPM/OPS). For this thesaurus, relations (temporal, causal, and intentional) and rule collections were essential.

Xmed was tested on millions of data sets of an anonymous comparison initiated by the German Medical Association and showed the best results. Three systems were compared, two commercial ones and Xmed. The judge, Dr. Graubner, did not know which system was used for the classification.

Xmed was capable to form a standardized meta-medical record from the mix of coded and free text data. This was the basis for the research cycle of the BAIK-Model. (Since my retirement in 2003 it has not been maintained anymore. However, it could be reactivated with moderate effort.)