Plenge et al.,
Family history mentions smoking and that seems to create some false positives. Also the negex negation algorithm missed a few negations. Discussed alternative approaches elaborated in the i2b2 NLP smoking bakeoff.
Also reviewed cardiovascular risk in the medical record and the gross lack of specificity of myocardial infarction obtained from claims/billing data as compared to the narrative text.
No comments:
Post a Comment