Cardiac tutorial #233
Cardiac tutorial #233
Conversation
- Based on Max's pipeline scripts - Includes sampling, simulation, emulation, and sensitivity analysis
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
Remaining comments which cannot be posted as a review comment to avoid GitHub Rate Limit
blackfmt
[blackfmt] reported by reviewdog 🐶
[blackfmt] reported by reviewdog 🐶
autoemulate/autoemulate/simulations/flow_functions.py
Lines 138 to 149 in 0f1382d
Coverage report
Click to see where and how coverage changed
This report was generated by python-coverage-comment-action |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #233 +/- ##
==========================================
- Coverage 94.17% 89.41% -4.77%
==========================================
Files 62 64 +2
Lines 3606 3798 +192
==========================================
Hits 3396 3396
- Misses 210 402 +192 ☔ View full report in Codecov by Sentry. |
|
@MaxBalmus I managed to transfer our original commits into the autoemulate repo, I think it makes more sense to work here directly, so that Martin can chip in where appropriate. |
@MaxBalmus @aranas: This is great! It would be best to make the simulation a function to be consistent with the other simulations eventually, but this isn't urgent. Let me know when I should have a look at it! |
autoemulate/sensitivity_analysis.py
Outdated
|
|
||
| # Use custom names if provided, else default to "x1", "x2", etc. | ||
| parameter_names = ( | ||
| problem["names"] if problem is not None else [f"x{i+1}" for i in range(len(next(iter(results.values()))["S1"]))] |
There was a problem hiding this comment.
[blackfmt] reported by reviewdog 🐶
| problem["names"] if problem is not None else [f"x{i+1}" for i in range(len(next(iter(results.values()))["S1"]))] | |
| problem["names"] | |
| if problem is not None | |
| else [f"x{i+1}" for i in range(len(next(iter(results.values()))["S1"]))] |
There was a problem hiding this comment.
This should resolve #279
marjanfamili
force-pushed
the
cardiac-tutorial
branch
from
34eb66e to
a36f768
Compare
autoemulate/metrics.py
Outdated
| discrepancy = np.atleast_1d(discrepancy) | ||
| n_obs = len(obs_mean) | ||
| rank = min(max(rank, 0), n_obs - 1) | ||
| # Vs represents the total variance associated with the observations, predictions, and potential discrepancies. |
There was a problem hiding this comment.
[blackfmt] reported by reviewdog 🐶
| # Vs represents the total variance associated with the observations, predictions, and potential discrepancies. | |
| # Vs represents the total variance associated with the observations, predictions, and potential discrepancies. |
…n benefit from removing some of teh hardcoded parameters and making them input.
There was a problem hiding this comment.
I made some suggestions, mostly around simplifying the notebook to make it just focus on autoemulate capability and nothing. In brief:
- the intro should just say the purpose of this tutorial is to demonstrate how autoemulate can be used to do sensitivity analysis (and maybe also to demonstrate how it works on a more complex example)
- I would also remove the end part which uses the SALib package to do some further sensitivity analyses
| @@ -0,0 +1,304 @@ | |||
| import matplotlib.pyplot as plt | |||
There was a problem hiding this comment.
Should this file be removed? It's a duplicate of a same named file in the autoemulate directory (where it should be).
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "## Context\n", |
There was a problem hiding this comment.
I'd reorder and slightly reword the intro section. Specifically, I'd remove context (people who are reading the autoemulate docs already know this) and start with purpose of the tutorial instead i.e., explain why someone reading the docs should read this page. Some of the context text explain gin the cardio side might need to be moved elsewhere.
I made some rough edit suggestions to this end but they are not necessarily comprehensive.
| "## Purpose\n", | ||
| "In this tutorial, we will demonstrate the application of `AutoEmulate` within a cardiac modeling pipeline, specifically focusing on the reconstruction of critical heart parameters using synthetic data. The goal is to demonstrate how these measures—such as arterial systemic pressure, and flow rates —can be leveraged to estimate a range of model parameters related to cardiac mechanics and cardiovascular hemodynamics. \n", | ||
| "\n", | ||
| "By the end of this tutorial, participants will have a clear understanding of how to use emulation to facilitate the real-time simulation of complex cardiac dynamics on a local CPU.\n", |
There was a problem hiding this comment.
Using the word participants makes this sound like it's meant for an external event. I'd drop that throughout and instead use the style of the rest of the documentation ("we will do this...").
| "By the end of this tutorial, participants will have a clear understanding of how to use emulation to facilitate the real-time simulation of complex cardiac dynamics on a local CPU.\n", |
| "\n", | ||
| "By the end of this tutorial, participants will have a clear understanding of how to use emulation to facilitate the real-time simulation of complex cardiac dynamics on a local CPU.\n", | ||
| "\n", | ||
| "Participants will learn how to explore the relationship between underlying cardiac health parameters and non-invasive measures using synthetic data, with a focus on how these insights can improve understanding and estimation of key cardiovascular dynamics." |
There was a problem hiding this comment.
Similar to above, it feels like the purpose here is to learn about autoemulate, not cardiac health
| "Participants will learn how to explore the relationship between underlying cardiac health parameters and non-invasive measures using synthetic data, with a focus on how these insights can improve understanding and estimation of key cardiovascular dynamics." |
| "metadata": {}, | ||
| "outputs": [], | ||
| "source": [ | ||
| "import os\n", |
There was a problem hiding this comment.
I don't think this is used throughout
| "import os\n", |
| "1. setting up a simulation : The system simulated is a tube with an input flow rate at any given time. The tube is divided to 10 compartments which allows for the study of the pressure and flow rate at various points in the tube. \n", | ||
| "2. Running the simulation for 60 sets of parameters sampled from the parameter space. \n", | ||
| "3. applying Autoemulate to find the best emulator fot this simulation \n", | ||
| "5. assessing model accuracy, performing sensitivity analysis, and comparing computational resources for different modeling approaches. " |
There was a problem hiding this comment.
| "5. assessing model accuracy, performing sensitivity analysis, and comparing computational resources for different modeling approaches. " | |
| "5. Assessing model accuracy and performing sensitivity analysis. " |
| ] | ||
| }, | ||
| { | ||
| "cell_type": "code", |
There was a problem hiding this comment.
I'm a bit confused by everything that follows from here onwards. The above cell demonstrates how to do sensitivity analysis with AutoEmulate but now it seems that we've reverted to using the SALib package to do some further custom analysis? I'd remove the rest of this section and if there is anything here that we think that autoemulate should capture we can add it to an issue to tackle separately.
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "# Scaling Sensitivity Analysis with Emulation" |
There was a problem hiding this comment.
This could either say "scaling sensitivity analysis with autoemulate" or I'd leave just as is given the user is coming across this in the autoemulate docs
| "# Scaling Sensitivity Analysis with Emulation" | |
| "# Scaling Sensitivity Analysis" |
| "metadata": {}, | ||
| "source": [ | ||
| "## Context\n", | ||
| "In the field of cardiovascular modeling, capturing the dynamics of blood flow and the associated pressures and volumes within the vascular system is crucial for understanding heart function and disease. Physics-based models that accurately represent these dynamics often require significant computational resources, making them challenging to apply in large-scale or real-time scenarios. Emulation techniques provide a way to achieve high-fidelity simulations of the cardiovascular system, allowing for efficient and accurate analysis of key hemodynamic parameters." |
There was a problem hiding this comment.
| "In the field of cardiovascular modeling, capturing the dynamics of blood flow and the associated pressures and volumes within the vascular system is crucial for understanding heart function and disease. Physics-based models that accurately represent these dynamics often require significant computational resources, making them challenging to apply in large-scale or real-time scenarios. Emulation techniques provide a way to achieve high-fidelity simulations of the cardiovascular system, allowing for efficient and accurate analysis of key hemodynamic parameters." |
| "metadata": {}, | ||
| "source": [ | ||
| "## Purpose\n", | ||
| "In this tutorial, we will demonstrate the application of `AutoEmulate` within a cardiac modeling pipeline, specifically focusing on the reconstruction of critical heart parameters using synthetic data. The goal is to demonstrate how these measures—such as arterial systemic pressure, and flow rates —can be leveraged to estimate a range of model parameters related to cardiac mechanics and cardiovascular hemodynamics. \n", |
There was a problem hiding this comment.
| "In this tutorial, we will demonstrate the application of `AutoEmulate` within a cardiac modeling pipeline, specifically focusing on the reconstruction of critical heart parameters using synthetic data. The goal is to demonstrate how these measures—such as arterial systemic pressure, and flow rates —can be leveraged to estimate a range of model parameters related to cardiac mechanics and cardiovascular hemodynamics. \n", | |
| "In this tutorial, we will demonstrate the application of `AutoEmulate` within a cardiac modeling pipeline, specifically focusing on the reconstruction of critical heart parameters using synthetic data. \n", |
this PR is a work in progress on the tutorial showcasing how autoemulate can be embedded into an end-to-end cardiac modelling pipeline including sensitivity analysis.
This tutorial still requires some improvements