Research on early prediction of cardiovascular and cerebrovascular diseases based on the CNN-xLSTM-CBAM model

Cardiovascular diseases (CVDs), due to their complex causes and diverse classifications, pose challenges for traditional machine learning methods, which often need more classification efficiency and accuracy when addressing these issues. This study proposes a convolutional neural network (CNN) model integrated with CBAM to overcome this limitation. The model incorporates an improved variant of LSTM, enhancing feature recognition capability and computational efficiency when processing complex cardiovascular disease data to achieve accurate early-stage CVD prediction. Using relevant medical datasets to establish prediction targets, the model employs spatial and channel attention mechanisms to enhance feature representation and predict coronary heart disease and stroke separately. Experimental results show that the model achieves prediction accuracies of 95.00% for coronary heart disease and 93.29% for stroke, significantly improving predictive performance and demonstrating the effectiveness and accuracy of this approach for complex disease prediction.