Extraction of the final diagnosis from medical treatment record based on deep-learning and An Aparatus Thereof

摘要

The present invention provides a method for extracting a final diagnosis from a medical readout record made by a computer, the method comprising: generating sequence data by representing a readout record showing a diagnosis history of a patient in the form of a word vector sequence; starting machining learning of the sequence data based on a convolutional neural network (CNN) and a recurrent neural network (RNN); and acquiring a final diagnosis using the data processed by machine learning. The present invention provides a solution to extract the disease name of the final diagnosis which can be applied to all different formats of readout records in different hospitals. This may increase the efficiency of analysis of the readout records. A final diagnosis inference network of the present invention in which the CNN and the RNN are combined overcomes the limitations on data format and sentence length, which are disadvantages of a text reference network based only on conventional CNN, thereby showing high performance.