Evaluating closed-loop effects from vasodilator administration for pulmonary hypertension treatment

Agnieszka Borowska, Mitchel J. Colebank, Mette S. Olufsen, Dirk HusmeierProceedings of the International Conference on Statistics, Theory and Applications (ICSTA ...) 2023 · DOI

Abstract

Closed-loop data science is emerging as a critical tool for clinical decision-making, particularly in scenarios where medical interventions alter the underlying physiology a model was originally designed to predict. In the treatment of Pulmonary Hypertension (PH), clinicians often face a "data decay" problem: high-resolution, invasive data is available at diagnosis, but follow-up monitoring is typically limited to non-invasive, less granular tests. This paper explores a mathematical framework using a 1D fluid dynamic model to evaluate these closed-loop effects. By replacing simplified Windkessel electrical circuit analogies with more physiologically realistic structured tree boundary conditions, the research aims to better characterize how vasodilators impact the pulmonary arterial network. While this transition introduces complexities in parameter identifiability, it provides a necessary foundation for robust clinical decision support.

Introduction: The Challenge of Closed-Loop Dynamics

In many data science applications, the system under observation is assumed to be static or independent of the observer's actions. However, in clinical medicine, the act of treatment creates a "closed-loop" effect. When a patient with Pulmonary Hypertension (PH)—a condition characterized by dangerously high blood pressure in the lungs—is prescribed vasodilators, the medication dilates the pulmonary arteries to reduce pressure. This intervention fundamentally changes the patient's physiology.

The challenge for clinicians is that while the initial diagnosis is supported by a comprehensive suite of tests (often invasive), follow-up visits rely on non-invasive, less expensive monitoring. Consequently, the original calibrated model of the patient's cardiovascular system becomes outdated, and there is insufficient new data to fully re-calibrate it. This research addresses the need for models that can account for these physiological shifts without requiring continuous invasive testing.

Methodology: From Circuits to Trees

The researchers utilize a 1D fluid dynamic mathematical model to predict blood pressure, flow, and vessel area within the pulmonary arterial network. The primary innovation lies in the refinement of the model's boundary conditions, which represent the resistance and elasticity of the smaller, downstream blood vessels.

The Evolution of Boundary Conditions

Historically, these models utilized Windkessel elements. These are analogies to electrical circuits (using resistors and capacitors) that represent the cumulative resistance and compliance of the vasculature. While computationally efficient, Windkessel models are essentially "black boxes" that lack anatomical detail, making it difficult to pinpoint exactly how a drug is affecting specific regions of the lung.

To improve physiological realism, the authors have replaced these circuit analogies with a structured tree model. This approach simulates a detailed, branching network of smaller vessels. By modeling the vasculature as a tree rather than a simple circuit, the framework can more accurately reflect the physical dilation caused by vasodilators across the arterial network.

The Trade-off: Realism vs. Identifiability

Moving from Windkessel elements to a structured tree model introduces a classic modeling dilemma: the trade-off between realism and identifiability.

  1. Increased Realism: The structured tree model provides a more granular look at how medication changes the physical properties of the lung's blood vessels.
  2. The Identifiability Hurdle: As the model becomes more complex, the number of parameters increases. "Parameter identifiability" refers to the ability to uniquely determine these internal variables from available clinical data. If multiple different physiological states produce the same non-invasive test results, the model cannot provide reliable guidance for clinical decisions.

Conclusion and Future Directions

This research represents a significant methodological shift in cardiovascular modeling. By integrating structured tree models into a 1D fluid dynamics framework, the authors have established a more realistic basis for tracking the effects of PH treatments.

While the transition introduces significant mathematical hurdles regarding parameter interpretation, it is a necessary step toward "closed-loop data science" in the clinic. Future work will focus on developing statistical methods to ensure these complex parameters can be reliably inferred from limited follow-up data, eventually allowing doctors to predict a specific patient's response to vasodilator dosages with high precision.

Summary updated

Abstract

Closed-loop data science can facilitate clinical decision-making by addressing a common situation in which acting on predictions from a calibrated model affects the initial assumptions of the model, e.g. the patient's physiology, however the model cannot be re-calibrated. Consider the following, typical scenario. At an initial visit a patient undergoes a number of tests, some of which might be invasive, which allows clinicians to diagnose the patient and administer suitable medication. Monitoring at follow-up visits is typically limited to a reduced number of tests, preferably noninvasive and/or less expensive. The challenge is then to assess the impact the administered medications have had on the patient's health and physiology, i.e. the closed-loop effects that occur after human intervention. In the present paper we focus on evaluating closed-loop effects from vasodilator administration for treating pulmonary hypertension (PH), i.e. high blood pressure in the lungs. PH is a serious and potentially fatal disorder, typically treated by administrating vasodilators, i.e. drugs that dilate pulmonary arteries, thus causing the pulmonary arterial pressure (PAP) to decrease. In our framework we apply a 1D fluid dynamic mathematical model with structured tree boundary condition [1] The model predicts pressure, flow and vessel area for a given pulmonary arterial network and parameters characterising the patient's physiology. An earlier proof-of-concept study [2] has demonstrated the importance of closed-loop data science in cardio-vascular modelling. In that study, the detailed structure of the vasculature related to small blood vessels was approximated by electrical circuit elements (so-called ``Windkessel elements") to simplify the mathematical complexity. To make our model physiologically more realistic, we have now relaxed this assumption and replaced the Windkessel elements by a more detailed model of the downstream vasculature, i.e. the structured tree model [1]. This leads to challenging problems related to parameter identifiability and interpretability, and we will discuss how they can be addressed so that the model can be calibrated for clinical decision support applications.
BibTeX
@inproceedings{borowska2023evaluating,
  author = {Agnieszka Borowska and Mitchel J. Colebank and Mette S. Olufsen and Dirk Husmeier},
  title = {{Evaluating closed-loop effects from vasodilator administration for pulmonary hypertension treatment}},
  booktitle = {Proceedings of the International Conference on Statistics, Theory and Applications (ICSTA ...)},
  year = {2023},
  doi = {10.11159/icsta23.162},
}
Figure