Analysis, Recognition and Encoding of medical diagnosis data

As a result of legal provisions and present used clearing systems with hospital cost units in europe the field of medical diagnosis will gain more and more importance. On the one hand the diagnosis data in patient's files have to be complete and without any error, and on the other hand the costs of treatments have to be compensated by social insurance units using these medical diagnosis data. Processing the huge amount of diagnosis data, which is passed to social insurance units each year, requires adequate numerical encoding systems to determine expense of treatments. In most cases the medical coding standard ICD (International Classification of Deseases) is used, although originally it has not been designed for this purpose. Some hospitals use their own proprietary encoding systems and translate the results into ICD codes. No difference which systems hospitals use, the common way translating medical diagnoses into encoding systems is doing it manually and wasting precious time of qualified medical professionals. There are various reasons why computers are rarely used in this application domain. Medical diagnoses are written in natural language which means that every hospital has its own rules for making diagnoses and choosing the language (the countries native language, english or latin). Even in Austria there are some hospitals using latin for medical diagnoses, whereas most hospitals use german. The multilingual analysis of natural text is a growing field in medical informatics but the research is in its infancy. Because of the fact that even in one and the same hospital doctors write down the diagnoses of deseases and injuries differently, it seems to be rather difficult to design computer systems for translating them into encoding systems fully automatically.