Physiology-Based Identifiable Model-Quantity Subset Selection for Early Prediction of Intradialytic MAP Trajectories...

Area of Science:

• Biomedical Engineering
• Physiological Modeling
• Computational Fluid Dynamics

Background:

• Intradialytic hypotension (IDH) is a frequent and serious complication during hemodialysis, linked to mean arterial pressure (MAP) fluctuations.
• Predicting intradialytic MAP is challenging due to limited early data and non-identifiable parameters in physiological models.

Purpose of the Study:

• To develop a dimensionality-reduced predictive framework for early intradialytic MAP trajectory forecasting.
• To identify key physiological parameters for accurate MAP prediction using sparse observations.

Main Methods:

• A screen-select-estimate workflow was employed to identify crucial model quantities.
• Sensitivity analysis and Fisher Information Matrix (FIM) were used to select initial blood volume (Vb0) and post-capillary venous resistance (Rs3).
• A reduced model framework was validated using MAP data from the first 100 minutes of dialysis.

Main Results:

• The selected two-parameter model (Vb0, Rs3) effectively predicted MAP trajectories across four patient types.
• Model-derived terminal specific blood volume and RBV showed distinct ranges for different phenotypes.
• Prediction accuracy demonstrated low root-mean-square error (RMSE) and maximum error values.

Conclusions:

• The developed framework enables interpretable early prediction of MAP decline during hemodialysis.
• This approach provides a foundation for patient-specific evaluation of IDH-related MAP changes.
• The study highlights the importance of Vb0 and Rs3 in understanding intradialytic hemodynamic variability.