Fig 1.
Abbreviations: ADMIN, administrative data, AE, adverse event, AL, algorithm, AUC, area under the receiver operating characteristic curve, EMR, electronic medical record, NPV, negative predictive value, PHN, personal health number, PPV, positive predictive value, Sen, sensitivity, Spe, specificity.
Table 1.
The adverse events list.
Fig 2.
The process of text classification for pressure ulcers (PU).
This picture shows the steps of text classification. We divide raw documents into sentences. All the words of each sentence are passed to the transformer to obtain a vector that can carry the sentence’s meaning. The vectors of all sentences are passed to another transformer to get the document embedding. The transformer can be seen as a "summarizer," step by step, the transformer abstract the meaning of the article related to the classification problem into the final document embedding.