[LSTM-XGBoost Based RR Intervals Time Series Prediction Method in Hypertensive Patients]

Area of Science:

• Cardiovascular physiology
• Machine learning in healthcare
• Biomedical signal processing

Context:

• Hypertension significantly impacts cardiac function and necessitates continuous patient monitoring.
• Accurate prediction of RR intervals is crucial for assessing heart conditions in hypertensive individuals.
• Existing single-model approaches for RR interval prediction have limitations.

Purpose:

• To develop and evaluate a combined machine learning model for predicting RR intervals in hypertensive patients.
• To enhance the accuracy of RR interval prediction by integrating Long Short-Term Memory (LSTM) and Gradient Boosted Trees (XGBoost) models.
• To assess the clinical feasibility of the proposed integrated model for patient management.

Summary:

• This study utilized data from 8 hypertensive patients to predict RR intervals.
• A combined model, integrating LSTM and XGBoost via the inverse variance method, was employed to overcome single-model prediction limitations.
• The combined LSTM-XGBoost model demonstrated improved prediction accuracy compared to individual models.

Impact:

• The developed LSTM-XGBoost model offers a novel computational method for predicting RR intervals in hypertensive patients.
• This approach has the potential for improved clinical decision-making and patient risk stratification.
• The findings suggest a feasible tool for enhancing the analysis and early warning of heart conditions in hypertensive populations.