Self-attention based recurrent convolutional neural network for disease prediction using healthcare data.

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Natural Language Processing for Clinical Text Analysis

Background:

• Traditional methods struggle with unstructured clinical text for disease diagnosis.
• Analyzing large volumes of clinical data is crucial for early disease detection.
• Existing machine learning models face challenges in handling the complexity and nuances of clinical text.

Purpose of the Study:

• To develop a deep learning model for accurate disease diagnosis from clinical text.
• To address limitations of traditional methods in analyzing unstructured medical data.
• To improve the prediction of diseases like cerebral infarction using advanced neural networks.

Main Methods:

• Implementation of a self-attention-based Recurrent Convolutional Neural Network (RCNN) model.
• Utilizing bi-direction recurrent connections within convolution to learn semantic features.
• Integrating a self-attention mechanism to focus on essential features in clinical text.

Main Results:

• The proposed model achieved an accuracy of 95.71% on a real-world hospital dataset.
• Experimental results demonstrate superior performance compared to existing methods for cerebral infarction prediction.
• The model effectively identifies key features within clinical text for disease risk assessment.

Conclusions:

• A self-attention-based RCNN model offers enhanced prediction for cerebral infarction.
• The model demonstrates significant improvements in disease risk prediction accuracy.
• This approach holds potential for predicting other diseases using clinical text data.