Simulations in Healthcare & Medicine

Most work described on this page was done as part of my PhD research at The Johns Hopkins University where I contributed to the development of an efficient and versatile model for simulating aortic valve dynamics. The model was used for gaining fundamental biophysics insights, biomarker discovery, medical device design, and procedural planning in the context of transcatheter aortic valve replacement (TAVR). I also showcase research on ocular blast trauma prediction and mitigation, to which I contributed at the Indian Institute of Technology Bombay.

Fundamental Biophysics Insights

Computational modeling can provide unparalleled fundamental insight into complex biophysics which may not be possible using in vivo/ in vitro models. As part of my doctoral work, I tested changes in individual aortic valve (AV) leaflet mobility on downstream biomechanics to understand how aortic stenosis (AS) influences blood flow and consequently the forces on the ascending aorta wall.

Differences in blood flow patterns arising from healthy and dysfunctional AVs are illustrated in the alongside figure showing a comparison of vortex structures colored by axial velocity contours resulting from healthy tricuspid and partially fused bicuspid AVs. Flow features colored in red correspond to forward flow while those in blue represent reverse flow. The healthy valve results in largely forward, axisymmetric flow with little reverse flow close to the aortic sinus, while the bicuspid valve leads to a deflected, non-circular jet which impacts the aorta lumen and results in a large recirculation zone shown in blue. These differences in hemodynamic characteristics can have severe implications for the health of the proximal aorta and are described in detail here.

Visualization of vortex structures and streamlines, colored using contours of axial (streamwise) velocity, for flow through (left) healthy tricuspid and (right) type-1 bicuspid aortic valves, during the deceleration phase of ventricular systole. This work was featured in UT Austin’s Texascale magazine.

Biomarker Discovery & Medical Device Development

Digital Auscultation-based Heart Valve Monitoring:

Cardiac auscultation is the practice of listening to heart sounds to make a medical diagnosis and has historically been used to detect a wide range of heart valve diseases (HVDs). It is advantageous in that it is inexpensive, non-invasive and can be performed anywhere but it generally suffers from poor diagnostic accuracy. Recently, AI-based digital stethoscopes have been developed to augment the sensitivity and specificity of cardiac auscultation, but developing such AI models requires large amounts of manually labeled time-series data, which are difficult to develop.

Computational modeling can address this challenge by synthesizing arbitrarily large databases of heart sounds resulting due to physiological and pathological valve function, which can be precisely controlled. In a paper I published in Frontiers in Physiology, I show how the fluid-dynamics associated with healthy and stenotic aortic valves result in characteristic sounds (listen to the sounds!) familiar to physicians. I also presented a supervised learning algorithm which can recognize these characteristics and enhance the accuracy of auscultation-based diagnosis.

Sensorized Transcatheter Aortic Valves:

Transcatheter prosthetic heart valves (right) offer a safe, effective and minimally-invasive option for valve replacement. But these valves are susceptible to premature failure in rare cases due to thrombus buildup on their surface which can lead to potentially fatal outcomes. In early stages, valve dysfunction may be asymptomatic and the recipient may not manifest signs of distress until considerable valve degradation has already occurred. Real-time knowledge of valve function enables catching potential valve failure in its early stages and prevent serious complications due to valve failure.

I laid the groundwork for a “sensorized” prosthetic valve (see Figure alongside) with passive pressure sensors embedded at strategic locations along its frame. I used the simulations described above to extract signature hemodynamic responses of healthy and stenotic valves. I trained a data-driven algorithm which analyzes these signatures and accurately predicts the presence of reduced mobility in the valve leaflets even in early stages. This work is still in progress, and I published my findings in two papers and contributed to a patent. Further development and eventual translation of this technology would enable early detection of valve failure and patient-tailored anti-thrombotic therapy.

You can also learn more about this work from my presentation at 73rd annual meeting of the American Physical Society’s Division of Fluid Dynamics conference:

Procedural Planning/ Surgical Guidance

Injury Prediction

Since computational modeling can estimate biomechanical trauma, simulation results can be analyzed to predict injury risk and aid in the design of trauma mitigation strategies. For example, the above figure shows the impact of explosive overpressures generated by an air blast, on the human head. Scenarios with no ocular protection (above, left) and with protection using tactical eye armor (above, right; armor hidden for visibility) were simulated. Facial features cause the blast wave to focus on the bare eye, leading to large biomehcanical loading. One the other hand, protective eye armor bears the major brunt of the overpressure and prevents blast wave focusing, leading to considerable mitigation in ocular loading. You can learn more about these simulations in this paper. Damage incurred by individual ocular tissues can be estimated using finite-element models of the eye (right). A later paper details the biomechanical response of important ocular components to blast loading and the injury risk estimated for the bare and protected eyes. Such simulations can provide insights into mechanical pathways for disease and aid in the design of protective gear for targeted mitigation in tissue damage.