Biomedical Informatics 4th Edition by Edward Shortliffe, James Cimino

Exercise 6

In the two following scenarios, an outof-the-shelf NLP system that identifies terms and normalizes them against UMLS concepts, is applied to a large corpus of texts. In the first scenario, the corpus consists of patient notes. Looking at the frequency of different concepts, you notice that there is a large number of patients with the concept C0019682 (HIV) present, much larger than the regular incidence of HIV in the population reported in the literature. In the second scenario, the corpus consists of full-text biology articles published in PubMEDCentral. Looking at the frequency of different concepts, you notice that the failed axon connection (fax) gene is one of the most frequently mentioned genes in your corpus. Describe how you would check the validity of these results. For both cases, discuss what could explain the high frequency counts.