Processing Text in Medical Databases

摘要: In the 1950s clinical data in medical records of patients United States were mostly recorded a natural, English-language, textual form. This was commonly done by physicians when recording their notes on paper sheets for patient’s history and physical examination, reporting interpretations x-ray images electrocardiograms, dictated descriptions surgical procedures. Such patients’ generally health-care professionals as hand-written notes, or reports that then transcribed typed sheets, all collated paper-based charts; these charts stored shelves record room. The process manually retrieving from always cumbersome time consuming. An additional frequent problem patient seeing more than one physician same day facility; chart often left first doctor’s office, therefore not available to other who had see without having access any prior information. Pratt (1974) observed professional collected during care largely non-numeric form, formulated almost exclusively English language. He noted word, phrase, sentence this language understood spoken read; marks punctuation order presentation words represented quasi-formal structures could be analyzed content according common rules for: (a) recognition validation string matter morphology syntax; (b) registration each datum its meaning semantics; (c) mapping recognized, defined, syntactical semantic elements into structure reflected informational original string, (d) processes required definition interpretation information user.