Taking the pulse on less computationally intensive cardiac digital twins

Digital twins — precise virtual replicas of physical systems updated with real-time data — are a compelling tool for healthcare, as doctors can noninvasively model patients’ conditions and potential interventions. But cardiac simulations are incredibly computationally intensive, requiring updating many millions of fluid points for a single heartbeat. To reduce the computational load, Khan et al. streamlined the calibration process for virtual cardiac cycles in the coronary artery.

While the standard practice after a change in heart rate is to wait five or more simulated cardiac cycles to achieve stable data, the team found that only one to two cycles were needed. The result should lead to faster, less computationally expensive data.

“Only then can we efficiently build a person’s hemodynamic map over longer periods like weeks or months,” said author Nusrat Sadia Khan.

Since large changes in a patient’s heart rate are a sticking point for cardiac digital twins, the team modeled the largest reasonable swing: 50 to 200 bpm. They also modeled parameters like fluid velocity, pressure, and shear stress. In extreme and realistic cases, they assessed how many full previous cardiac cycles in the coronary artery were needed to verify that the next beat was returning stable data: always two or fewer.

Next, the team wants to test more sections of the heart, like the pulmonary and carotid arteries. Due to the high volume of heart disease patients, they hope digital twin technology will bring care from a reactive to proactive approach.

“This study provides the foundation for making digital twins stable across realistic heart rate changes, which is a key step toward using them for long-term patient care,” said Khan.

Source: “Establishing hemodynamic convergence framework for coronary digital twins under realistic dynamic heart rates,” by Nusrat Sadia Khan, Cyrus Tanade, Justen Geddes, and Amanda Randles, Physics of Fluids (2025). The article can be accessed at https://doi.org/10.1063/5.0287796 .