My research focuses on the computational modeling of the heart. During my PhD, I have developed a number of tools to better understand the function of the heart in healthy and diseased states:

Predicting drug-induced arrhythmias by multi-scale modeling 

Screen Shot 2018-10-17 at 10.27.46 PM.png

Drugs of any kind can have adverse effects on the heart. Torsades de Pointes is a particular kind of arrhythmia caused by drugs, which can quickly degenerate into lethal ventricular fibrillation. In order to design safe drugs, it is critical for pharmaceutical companies to  identify at early stages which compounds will cause this kind of arrhythmia. The current bio-markers used for this purpose are sensitive, but non-specific, leaving potentially life-saving drugs out of the development process. In this project, we developed a mechanistic, high-resolution, multi-scale model of the heart to predict drug-induced arrhythmias. Our model spontaneously develops Torsades de Pointes when high risk drugs are applied and maintains a normal rhythm when low risk drugs are applied. This tool will provide invaluable insights to drug developers and will accelerate the design of safer drugs. This model also presented a series of technical challenges due to high level of detail at cellular scale. To study each drug, we performed 1,000,000 time steps updating 250,000,000 variables. We partnered with SIMULIA, Uber Cloud, HPE and Advania to harness the advantages of a cloud-based high performance computer system. The result of this collaboration earned us the 2017 Hyperion Research Innovation Excellence Award.

Sahli Costabal, F., Yao, J., & Kuhl, E. (2018). Predicting drug‐induced arrhythmias by multiscale modeling. International journal for numerical methods in biomedical engineering34(5), e2964.

Sahli Costabal, F., Yao, J., & Kuhl, E. (2018). Predicting the cardiac toxicity of drugs using a novel multiscale exposure-response simulator. Computer Methods in Biomechanics and Biomedical Engineering, 21, 232–246.

Sahli Costabal, F., Sher, A., Yao, J., & Kuhl, E. (2018). Predicting critical drug concentrations and torsadogenic risk using a multiscale exposure-response simulator. Progress in Biophysics and Molecular Biology, accepted.

In the news: HPCwireInsideHPC

Generating Purkinje networks in the human heart

Screen Shot 2018-10-17 at 10.28.48 PM.png

The Purkinje network is a key element in the heart’s electrical system. It is responsible for the fast propagation of the electrical in the ventricles to allow a coordinated action of the muscle to pump blood efficiently. However, it is not possible to capture the structure of network with any in vivoimaging modality. This harms the ability to create personalized computational models of the heart to study individual patients and diseases. In this work, we developed an algorithm to create a representation of the Purkinje network on an arbitrary heart model. We went one step further and showed that the networks generated by this algorithm reproduced the key features of healthy and diseased electrocardiograms.

The code from this paper is open source and you can find it here:

Sahli Costabal, F., Hurtado, D. E., & Kuhl, E. (2016). Generating Purkinje networks in the human heart. Journal of biomechanics49(12), 2455-2465.

The role of mechanics in cardiac arrhythmias

Although cardiac arrhythmias are a primarily electrical phenomenon, the motion of the heart can influence the behavior of these abnormal rhythms. The most relevant driver of this coupling is called mechano-electrical feedback. Here, when cells are stretched, ion channels open and release current. Additionally, the domain in which the electrical wave is propagated deforms as the cardiac muscle contracts. In this work, we studied, from a modeling perspective, how critical electrophysiological metrics are affected by this electro-mechanical coupling. We found that the conduction velocity of the electrical wave and the spiral wave trajectory is affected by the level of detail included in the model. This work quantified the influence of certain assumptions that modelers do in this field, allowing them to make informed decisions.

Sahli Costabal, F., Concha, F. A., Hurtado, D. E., & Kuhl, E. (2017). The importance of mechano-electrical feedback and inertia in cardiac electromechanics. Computer methods in applied mechanics and engineering320, 352-368.

Understanding arrhythmias from complex electrical signals

Atrial fibrillation is the most prevalent kind of arrhythmia and it affects more than 6 million people in the US. In this project I studied a novel treatment for this disease. Here, the electrical waves in the atria are mapped with electrodes to determine to optimal location to apply the treatment. However, the signals coming from the electrodes are complex and noisy. In this work, we used computational modeling to better understand how these signals are generating from electrical waves in the diseased state. We showed that using traditional methods to determine the activation patterns can lead to erroneous interpretations. We corroborated this finding in successfully treated patients.:

Sahli Costabal, F., Zaman, J. A., Kuhl, E., & Narayan, S. M. (2018). Interpreting activation mapping of atrial fibrillation: A hybrid computational/physiological study. Annals of biomedical engineering46(2), 257-269.