A New Method for the Task of Automated ICD Coding Using Deep Learning

The aim of this research proposal is to investigate a deep learning method for automated ICD coding that builds upon existing research. The proposed method will focus on the integration of natural language processing (NLP) techniques, such as recurrent neural networks (RNN) and convolutional neural networks (CNN) with existing strategies for ICD coding. By leveraging the capability of deep learning with existing NLP techniques, the proposed method should be able to identify the text in an electronic health record (EHR) that are relevant to an ICD code, as well as to use the identified text to extract the proper ICD code. To enable the proposed method to accurately classify the text in EHR to a given ICD code, a large annotated corpus of EHR data will be required. This corpus should include EHR data that have been correctly coded with ICD codes, as well as the corresponding unstructured textual data. By leveraging this corpus, the proposed method will be able to learn how to accurately classify the text in new EHRs. In addition to NLP techniques, the proposed method will also include the integration of other cutting-edge approaches for ICD coding. These approaches include template-based algorithms and graph-based algorithms. By combining these strategies with NLP techniques, the proposed method will be able to identify relevant text for ICD coding and accurately classify the text to the corresponding ICD code. Finally, the proposed method will be evaluated on a test set of EHRs to measure its accuracy in identifying and assigning ICD codes. This evaluation will allow us to compare the performance of the proposed method with existing approaches and assess its efficacy. Overall, the proposed research will investigate a deep learning method for automated ICD coding that builds upon existing research. By integrating existing strategies with NLP techniques, the proposed method should be able to identify relevant text and classify the text to the corresponding ICD code with high accuracy. The efficacy of the proposed method will be tested on a test set of EHRs and compared to existing approaches.