This repository contains the supplementary material to the paper K.K. Harrod, J.L. Rogers, J.A. Feinstein, A.L. Marsden and D.E. Schiavazzi, Predictive modeling of secondary pulmonary hypertension in left ventricular diastolic dysfunction.
We provide the two datasets used to generate the results in the paper. Additionally we also provide some python scripts to run zero-dimensional hemodynamic models representing simple RC and RCR models, and one model for adult cardiovascular physiology.
Two datasets are provided for the adult model:
-
validation_dataset.csv includes anonymized hemodynamic data for a healthy patient and a patient with moderate and severe left ventricular diastolic dysfunction. See the paper for further details on this dataset.
-
EHR_dataset.csv includes anonymized hemodynamic data for 84 patients. See the paper for further details on this dataset.
Additionally, other two files are used by the non-autonomous RC and RCR models:
-
inlet.flow contains a time table wit the pulsatile inflow from the two models.
-
rc_dataset.csv contains two measurements for minimum, maximum and average pressure at the inlet of the two models.
-
rcModel - A model of a simple RC circuit subject to a time dependent inflow.
-
rcrModel - A model of a simple RCR circuit subject to a time dependent inflow.
-
lpnAdultModel - The circulation model used in the paper. Note how this model is already setup to assemble a Gaussian likelihood using the validation and EHR datasets provided in the data folder, as discussed in the paper.
The following libraries are required:
- numpy >= 1.19.2
- scipy.signal >= 1.5.2 - peek finding functionalities to compute acceletation/deceleration times in valve and E/A peak ratios for volumetric flow across valves.
- cython >= 0.29.19 - to reduce computation time RHS of the system of ODE and the RK4 time integration routines are implemented in cython.