AI-optimized GRU-based self-attention model for predictive diabetes staging in IoT healthcare 5.0

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Biomedical Engineering

Background:

• Healthcare 5.0 environments leverage Internet of Things (IoT) devices for user data collection.
• Early detection and staging of diabetes are crucial but challenging due to complex healthcare feature interrelationships.
• Accurate prediction of diabetes requires advanced analytical models.

Purpose of the Study:

• To develop and evaluate a novel self-attention Gated Recurrent Unit (GRU) model for predictive diabetes detection.
• To improve the accuracy of diabetes prediction by capturing temporal and spatial features in healthcare data.
• To compare the proposed model's performance against established deep learning baselines.

Main Methods:

• A GRU-based self-attention mechanism was employed to capture temporal dependencies and spatial features.
• Convolutional Neural Network (CNN) with Batch Normalization and ReLU was utilized for final classification.
• The model was trained and evaluated on healthcare datasets for diabetes prediction.

Main Results:

• The proposed self-attention GRU model achieved 93.94% accuracy, 95.28% precision, and 93.94% recall.
• The model demonstrated a high Area Under the Curve (AUC) of 0.9697.
• Performance significantly surpassed GRU, LSTM, RNN, and transformer-based baseline models.

Conclusions:

• The self-attention GRU model is effective for early and accurate diabetes detection in Healthcare 5.0.
• The model's ability to capture complex dependencies enhances predictive capabilities for chronic diseases.
• This approach offers a promising advancement in AI-driven diabetes management and prediction.