An Explainable Intelligent Approach for Cardiovascular Disease Diagnosis from High-Dimensional Laboratory Test Results

Cardiovascular diseases (CVD) constitute a group of chronic ailments that can have sudden onset, emphasizing the crucial need for their prediction and early prevention. The current clinical diagnosis of CVD often requires comprehensive examinations, including laboratory and imaging tests, symptom checks, and physical examinations. These procedures can be time-consuming, labor-intensive, and expensive for patients. Additionally, the utilization of accessible tabular laboratory test results, which are typically high-dimensional and imbalanced, presents substantial challenges for CVD diagnosis. To address these issues, we propose an explainable intelligent approach for CVD diagnosis by analyzing high-dimensional and imbalanced laboratory test results. Firstly, a conditional tabular generative adversarial network is introduced to address the issue of data imbalance. Furthermore, the proposed approach utilizes a one-dimensional convolutional neural network with self-attention mechanism to improve the accuracy and efficiency of CVD diagnosis. Moreover, by extracting the attention vectors, the importance of different features is analyzed to enhance the interpretability of the model. As a result, the proposed approach outperforms other methods by demonstrating improved prediction performance, achieving an accuracy rate of 88.89%. Furthermore, the most influential features that contribute to discovering the CVD diagnosis pattern are identified as risk factors, offering enhanced support for the accurate prevention of CVD.