Global sensitivity analysis informed model reduction and selection applied to a Valsalva maneuver model.

Area of Science:

• Physiology
• Computational Biology
• Biomedical Engineering

Background:

• The Valsalva maneuver (VM) significantly impacts cardiovascular regulation.
• Accurate modeling of heart rate response to VM requires understanding baroreceptor influences.
• Global sensitivity analysis (GSA) offers powerful tools for dissecting complex model behaviors.

Purpose of the Study:

• To develop and apply a GSA-informed methodology for model reduction and selection.
• To investigate the necessity of including both aortic and carotid baroreceptors in VM heart rate prediction models.
• To compare standard and novel time-varying GSA methods for parameter influence analysis.

Main Methods:

• Application of four GSA methods, including standard scalar and time-varying Sobol' indices (SIs).
• Introduction of a novel limited-memory SIs technique using a moving window approach.
• Model reduction comparing full, aortic-only, and carotid-only baroreceptor models, with selection via information criteria and neurological prediction comparison.

Main Results:

• Standard and time-varying GSA methods effectively quantified parameter influence on heart rate prediction during VM.
• Limited-memory SIs provided a dynamic assessment of parameter importance over time.
• Model reduction and selection revealed that both aortic and carotid baroreceptor regions are crucial for accurate VM simulation.

Conclusions:

• A GSA-informed approach facilitates effective model reduction and selection in physiological modeling.
• The aortic and carotid baroreceptor pathways are both essential components for simulating the heart rate response to the Valsalva maneuver.
• Time-varying GSA, particularly limited-memory SIs, offers valuable insights into dynamic parameter importance.