Visual Indicator for Intradialytic Hypotension Prediction Using Variation and Compensation of Heart Rate.

Area of Science:

• Nephrology
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Current intradialytic hypotension (IDH) prediction methods rely on baseline and ultrafiltration data, limiting real-time application.
• Real-time identification of patient condition during dialysis remains a challenge for existing IDH prediction technologies.

Purpose of the Study:

• To develop and validate a novel IDH indicator for real-time patient monitoring during dialysis.
• To utilize heart rate (HR) variability for immediate detection of intradialytic hypotension.

Main Methods:

• A new IDH indicator was developed, analyzing real-time average heart rate (HR) to infer blood pressure changes.
• Data from 40 dialysis patients were collected, focusing on HR changes in the 40 minutes preceding IDH events.
• A multilayer perceptron model was trained using HR data to predict IDH, with varying layer structures.

Main Results:

• The IDH indicator visually represents HR changes, enabling medical staff to respond promptly.
• The multilayer perceptron models achieved high accuracy in predicting IDH, with two-layer and five-layer models showing 88.6% and 85.2% accuracy, respectively.
• Analysis of upper and lower maximum HRs relative to the average HR formed the basis for prediction.

Conclusions:

• The combination of a visual IDH indicator and a multilayer perceptron model facilitates effective real-time management of intradialytic hypotension.
• This approach enhances medical staff's ability to respond to IDH events as they occur.
• The study demonstrates the potential of using HR dynamics for immediate and accurate IDH detection.