Artificial Intelligence Algorithm with ICD Coding Technology Guided by the Embedded Electronic Medical Record System...

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Clinical Documentation

Background:

• Accurate International Classification of Diseases (ICD) coding is crucial for healthcare statistics, billing, and research.
• Electronic Medical Record (EMR) systems present challenges in consistent and accurate ICD coding due to complex data and potential coder errors.
• Existing coding practices often suffer from inaccuracies in main diagnosis selection, disease classification, and handling of specific conditions or complications.

Purpose of the Study:

• To evaluate the application of ICD coding technology integrated with an EMR system.
• To investigate the optimization of medical record data using Natural Language Processing-Bidirectional Recurrent Neural Network (NLP-BIRNN) algorithms.
• To compare the performance of NLP-BIRNN against other neural network models for medical record analysis.

Main Methods:

• Established an EMR information knowledge system incorporating data from eight clinical departments.
• Collected patient medical records and disease diagnostic codes for statistical analysis.
• Employed a Natural Language Processing-Bidirectional Recurrent Neural Network (NLP-BIRNN) algorithm to process and optimize medical records, comparing its accuracy, symptom accuracy, and recall against Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN).

Main Results:

• Identified significant issues in current ICD coding practices, including unclear main diagnosis selection, incorrect disease classification, inaccurate coding of postoperative complications, incomplete diagnoses, and overly general code selection.
• The NLP-BIRNN algorithm demonstrated higher accuracy, symptom accuracy, and symptom recall compared to CNN and RNN models.
• AI-driven pathological language processing offers convenience for disease diagnosis and treatment.

Conclusions:

• Enhanced communication and knowledge training for coders and medical personnel are necessary to improve ICD coding quality.
• The NLP-BIRNN algorithm shows significant potential for optimizing medical records and improving the accuracy of ICD coding within EMR systems.
• Artificial intelligence holds promise for facilitating disease diagnosis and treatment through advanced language processing of clinical data.