A heart failure classification model from radial artery pulse wave using LSTM neural networks

Area of Science:

• Biomedical engineering
• Artificial intelligence in healthcare
• Cardiovascular diagnostics

Background:

• Heart failure (HF) is a major global health concern requiring accessible early detection methods.
• Deep learning (DL) offers promising non-invasive, rapid, and cost-effective solutions for HF detection.

Purpose of the Study:

• To evaluate the efficacy of deep learning algorithms in classifying heart failure (HF) using radial artery pulse wave data.
• To identify the most effective DL model for differentiating between healthy individuals, coronary artery disease (CAD) patients, and HF patients.

Main Methods:

• Utilized radial artery pulse wave data from 462 participants (healthy, CAD, HF).
• Applied and compared four DL models: LSTM, CNN, GRU, and Bi-LSTM after data preprocessing and balancing.
• Employed 10-fold cross-validation for robust performance and stability assessment.

Main Results:

• The Long Short-Term Memory (LSTM) model achieved the highest mean accuracy (0.8595 ± 0.0522).
• LSTM demonstrated superior performance across key metrics, including Area Under the Curve.
• SHAP framework was used for model interpretability and feature importance analysis.

Conclusions:

• An LSTM model analyzing radial artery pulse waves effectively distinguishes between healthy, CAD, and HF states.
• This non-invasive, cost-effective approach shows potential for early heart failure screening.
• Further clinical validation is recommended to confirm the tool's real-world applicability.