Deep-learning natural language processing for oncological applications

Yet, most medical information is stored in a free-text format. [...]the data are inaccessible for computer analysis and their potential remains unused. [...]the algorithm is tested with new information that it has not seen before and then uses patterns learned during training to analyse this new information. Another advantage of deep learning over traditional NLP methods is that deep learning models generally do not require previous human annotation of terms. [...]these models are less time consuming to develop and implement compared with traditional NLP methods.