Utilizing a Two-Stage Taguchi Method and Artificial Neural Network for the Precise Forecasting of Cardiovascular...

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Cardiovascular Disease Research

Background:

• Cardiovascular disease (CVD) onset is complex, necessitating early detection for effective prevention.
• Personal devices offer potential for point-of-care testing (POCT) to enhance CVD risk prediction.
• Existing prediction models often require significant computational resources, limiting their use on low-power devices.

Purpose of the Study:

• To introduce and evaluate a two-stage Taguchi optimization (TSTO) method for improving artificial neural network (ANN) model accuracy in CVD risk prediction.
• To minimize computational costs associated with optimizing ANN hyperparameters.
• To assess the feasibility of applying the optimized ANN model on personal devices for POCT.

Main Methods:

• A two-stage Taguchi optimization (TSTO) method was developed to identify optimal hyperparameter levels and settings for an ANN model.
• The TSTO method was applied to the Kaggle Cardiovascular Disease dataset using a personal computer.
• Hyperparameter tuning involved identifying optimal values for hidden layer, activation function, optimizer, learning rate, momentum rate, and hidden nodes.

Main Results:

• The TSTO method identified optimal ANN hyperparameters: hidden layer=4, activation=tanh, optimizer=SGD, learning rate=0.25, momentum rate=0.85, hidden nodes=10.
• This configuration achieved a state-of-the-art accuracy of 74.14% in predicting cardiovascular disease risk.
• The TSTO method reduced experimental requirements by a factor of 40.5 compared to traditional grid search, significantly conserving computational resources.

Conclusions:

• The proposed TSTO method effectively enhances ANN predictive accuracy for cardiovascular risk.
• The method significantly reduces computational load, making it suitable for resource-constrained environments and adaptable for low-power personal devices.
• This approach supports the advancement of point-of-care testing for cardiovascular disease prevention.