By Rachel Clipp
Modeling and simulation for the purpose of medical training is critical to improving patient outcomes and educating professional caregivers without exposing patients to unnecessary risk. Though a number of training modalities (e.g., manikins and virtual environments) are currently available in simulation centers, many rely on instructor-driven specifications of vital signs and physiologic behavior; these tactics can lead to inconsistent and inaccurate training algorithms.
Pulse Physiology Engine
Whole-body computational physiology models can fill this training modality gap by consistently and accurately calculating patient responses to illness, injury, and treatment. The Pulse Physiology Engine is an open-source solution that addresses the need for computational physiology models (see Figure 1). Pulse consists of lumped-parameter models for individual systems and differential equations that represent feedback mechanisms, pharmacokinetic and pharmacodynamic behavior, and other behaviors like diffusion. Users can explore injuries, diseases, and treatments by modifying the model parameters. One can also modify the circuits to include various types of equipment, such as ventilators. The engine assembles matrices to reflect completed circuits and calculates the unknowns via an Eigen library solver [2].
Figure 1. Overview of the Pulse Physiology Engine. Figure courtesy of Kitware.
An extensive verification and validation suite—which covers resting physiology and more than 100 disease, trauma, and treatment scenarios—supports our models. The suite includes a nightly build process and a system that compares published data with waveforms and maximum, minimum, and mean computed values. To do so, it uses custom developed tools that create comparison data, perform analysis, and generate plots and tables. These results are available in the Pulse documentation and are updated for each release.
COVID-19 Simulation
The medical community experienced a ventilator shortage in the early stages of the COVID-19 pandemic. However, the idea of ventilating multiple patients with a single ventilator raised significant safety concerns. Though this treatment method is a last resort, exploring its feasibility was critical at the time. We were able to rapidly configure the Pulse models to simulate and predict the outcomes of two patients who were sharing a single ventilator [4]. We tested a variety of patient configurations with different ventilator settings and computed the predicted patient outcome metrics. The results of these simulations demonstrate that lung compliance, oxygenation and oxygen saturation indexes, and end-tidal carbon dioxide levels of individual patients correlate most closely with patient outcomes. The patient outcomes were satisfactory when the lung compliance difference between two patients was less than 12 mL/cmH2O and the oxygen saturation index difference was less than 2 mmHg (see Figure 2).
Figure 2. Results from the Pulse COVID-19 simulated outcomes study. Figure courtesy of [4].
Hemorrhagic Shock
Recent improvements to the Pulse Physiology Engine include a hemorrhage model that more accurately mimics dynamic response through hemorrhagic shock. We are also now capable of prescribing a hemorrhage by coupling a certain compartment (region of the body — i.e., right leg) with a severity between 0 and 1, which allows users who are not computer science or clinical experts to apply reasonable hemorrhages to specific regions. This methodology scales the hemorrhage based on specific compartment flow; 1.0 indicates that 95 percent of compartment flow is redirected to a hemorrhage and 0.0 corresponds to 0 percent redirection. We preserved the existing ability to prescribe a hemorrhagic flow rate with a compartment. Users can also specify the hemorrhage as either internal or external, which determines whether the hemorrhage flows into the body or out to the environment.
Figure 3. Results from the Pulse hemorrhage study. Figure courtesy of Kitware.
We recently completed a study that compared the Pulse hemorrhage model to multiple other physiology engines using experiments that duplicated an existing approach [1]. Our study specified the three hemorrhage scenarios as mild, moderate, and severe based on the Advanced Trauma Life Support guidelines of 50, 100, and 500 mL/min, respectively. We used the standardized patient profile of a 33-year-old male who weighs 75 kg with a height of 175 cm. The parameters of interest were heart rate (beats/min), mean arterial pressure (mmHg), blood volume (mL), cardiac output (mL/min), stroke volume (mL), hemoglobin content (g), respiration rate (breaths/min), epinephrine serum concentration (pg/mL), and urine output (mL/min). We compared newly acquired Pulse results to previously collected data from
BioGears, Muse, and
HumMod, as well as literature data about hemorrhages [3].
We utilized experimental data for the mean arterial pressure and cardiac output, which comprise two significant parameters during hemorrhage response. Next, we compared the results of the four computational models with the literature data. Overall, we found that Muse and Pulse both provided the most validated responses to hemorrhage in terms of magnitude and response time (see Figure 3). Ultimately, however, Pulse outperformed the remaining engines by offering the most realistic model. Pulse’s impressive performance—coupled with the fact that it is freely available through an open-source license—made it the clear leader in this study.
More information about Kitware’s Pulse Physiology Engine is available online or via the Pulse newsletter. If you would like to feature your Pulse use case, please email us at [email protected].
Rachel Clipp presented this research during a minisymposium presentation at the 2021 SIAM Conference on Computational Science and Engineering, which took place virtually in March.
References
[1] Barnes, J.J., Kiberenge, K., Sweet, R., Keller, J., & Konia, M.R. (2020). Comparing hemorrhage in human physiology simulation tools: How they compare with expected human physiology and each other. Simul. Healthc., 15(5), 310-317.
[2] Bray, A., Webb, J.B., Enquobahrie, A., Vicory, J., Heneghan, J., Hubal, R., …, Clipp, R.B. (2019). Pulse physiology engine: An open-source software platform for computational modeling of human medical simulation. SN Compr. Clin. Med., 1, 362-377.
[3] Hall, J.E. (2020). Guyton and Hall Textbook of Medical Physiology (13th Ed.). Philadelphia, PA: Elsevier.
[4] Webb, J.B., Bray, A., Asare, P.K., Clipp, R.B., Mehta, Y.B., Penupolu, S., …, Polar, S.M. (2020). Computational simulation to assess patient safety of uncompensated COVID-19 two-patient ventilator sharing using the Pulse Physiology Engine. Plos One, 15(11), e0242532.
 |
Rachel Clipp is a technical leader on Kitware's Medical Computing Team. She worked to launch the Pulse Physiology Engine, which has been incorporated in numerous commercial and government-funded products and programs. |