NTT Research Develops Scalable Method to Accelerate Personalized Cardiovascular Bio Digital Twin Creation
NTT Research, Inc., the Silicon Valley-based research division of NTT, Inc., a global technology and business solutions provider, has announced a significant research advancement in personalized cardiovascular modeling. Scientists at the company’s Biomedical Engineering Innovation (MEI) Lab have developed a new method for rapidly and scalably creating personalized cardiovascular Bio Digital Twins. The approach is designed to accelerate the development of patient-specific virtual models of the heart and circulatory system while maintaining the physiological accuracy and interpretability required for potential medical applications.
NTT, the parent organization of NTT Research, serves more than 75% of the Fortune Global 100 companies and invests billions of dollars annually in research and development. Through its research initiatives, the company is exploring innovative technologies across several scientific and engineering fields, including healthcare, biomedical engineering, and advanced computing.
The latest development from the Biomedical Engineering Innovation Lab, formerly known as the NTT Research Medical and Health Informatics (MEI) Lab, focuses on addressing an important challenge in cardiovascular digital twin technology: the time and computational resources required to create accurate models for individual patients.
A cardiovascular Bio Digital Twin is a virtual representation of a person’s heart and blood vessels constructed using clinical measurements and physiological information. The technology aims to provide researchers and, potentially, healthcare professionals with a digital environment in which they can examine cardiovascular function and evaluate possible treatment approaches before applying them to patients.
The research introduces a method that uses precomputed virtual populations as Bayesian priors to accelerate the personalization process. By performing substantial computational work in advance, the approach could make it possible to generate individual cardiovascular models more efficiently without compromising the physiological foundations of the technology.
The findings have been published in the peer-reviewed journal Frontiers in Physiology in a paper titled “Accelerating Cardiovascular Digital Twin Creation Using Precomputed Virtual Populations as Bayesian Priors.”
Understanding Cardiovascular Bio Digital Twins
Cardiovascular Bio Digital Twins represent an emerging approach to personalized healthcare. Unlike conventional statistical models that primarily identify relationships and patterns within datasets, these models are designed to represent the underlying physiological mechanisms of an individual’s cardiovascular system.
A cardiovascular digital twin uses measurements such as blood pressure, heart rate, and stroke volume, which refers to the amount of blood pumped by the heart with each heartbeat. By combining these measurements within a physiological framework, researchers can construct a virtual model that reflects how an individual’s heart and circulatory system function.
The model can then be used to investigate how changes in one cardiovascular parameter influence other aspects of the system.
For example, when the amount of blood returning to the heart changes, the heart’s pumping response may also change. A physiologically based digital twin can simulate this relationship and demonstrate how the different components of the cardiovascular system interact.
This capability is important because cardiovascular function depends on multiple interconnected processes. Changes in blood volume, vascular resistance, heart rate, and cardiac performance can influence overall circulation.
A model that captures these relationships can provide more meaningful information than a system that simply predicts outcomes based on statistical associations.
Another important feature of a cardiovascular Bio Digital Twin is that its parameters correspond to actual physiological properties. This allows researchers to investigate not only what a model predicts but also the biological mechanisms responsible for the predicted outcome.
The model used in the latest NTT Research study is called Otto, named after German physiologist Otto Frank. His scientific contributions helped establish the understanding of how the heart adjusts the strength of its contractions in response to changes in the amount of blood entering it.
The Otto model provides the physiological foundation for the research team’s approach to creating personalized cardiovascular digital twins.
Precomputed Virtual Populations Support Faster Model Creation
One of the major challenges associated with cardiovascular digital twin development is the computational effort required to generate an accurate patient-specific model.
Traditional approaches may require extensive calculations to determine the physiological parameters that best represent an individual patient’s cardiovascular system. This process can become particularly demanding when multiple measurements and complex interactions need to be considered.
The NTT Research Biomedical Engineering Innovation Lab developed a strategy to address this challenge by creating a precomputed virtual population.
A virtual population is a collection of simulated individuals representing different combinations of physiological characteristics. Instead of constructing every model from the beginning when a new patient is evaluated, researchers can use an existing library of simulated cardiovascular systems as a starting point.
For the latest research, the team generated a collection of simulated healthy hearts and blood vessels using information from published physiological studies.
Each virtual individual in the population contains simulation outputs corresponding to approximately 20 interconnected cardiovascular measurements.
These measurements include important parameters such as blood pressure and the volume of blood pumped with each heartbeat. The values are designed to remain physiologically consistent with one another, meaning that they collectively describe a cardiovascular system capable of functioning realistically.
This interconnected structure is particularly important for developing useful digital twins. Cardiovascular measurements cannot always be evaluated independently because changes in one parameter may influence several others.
By creating virtual individuals with coordinated physiological measurements, the research team established a reusable library that can support the generation of personalized models.
Another important aspect of the approach is that the virtual population was developed using published physiological research rather than individual patient records. As a result, the library itself does not contain personal patient data.
This provides a scientific foundation for generating cardiovascular models without requiring the virtual population to be built from identifiable patient information.
Bayesian Priors Help Accelerate Personalization
The research paper describes how precomputed virtual populations can be used as Bayesian priors to accelerate the creation of cardiovascular digital twins.
Bayesian methods provide a mathematical framework for combining existing knowledge with new observations to estimate unknown parameters.
In the context of cardiovascular modeling, previously generated virtual individuals provide an informed starting point for identifying the physiological characteristics that may best represent a specific patient.
When clinical measurements become available, the model can use the precomputed population to guide the personalization process. Rather than performing all calculations from the beginning, the system can identify relevant physiological configurations from the existing library and use them to support the creation of an individualized model.
This strategy is intended to reduce computational requirements and shorten the time needed to generate patient-specific cardiovascular representations.
The approach also aims to preserve the physiological interpretability of the resulting models. This is important because healthcare applications require more than predictions alone. Researchers and physicians need to understand the physiological relationships behind the results.
By combining precomputed simulations with Bayesian inference, NTT Research is seeking to create a more efficient framework for developing personalized cardiovascular digital twins.
The ability to generate models rapidly could be particularly valuable in research environments involving large patient populations or clinical situations where timely information is important.
Potential Applications in Cardiovascular Disease Management
Cardiovascular diseases remain a major global healthcare challenge. According to the World Health Organization, cardiovascular diseases account for approximately 32% of deaths worldwide.
These conditions include heart failure, coronary artery disease, stroke, and other disorders affecting the heart and blood vessels.
Managing cardiovascular disease often requires physicians to consider multiple factors, including a patient’s medical history, physiological measurements, disease severity, and response to treatment.
However, individual patients may respond differently to the same medication or therapeutic intervention. Differences in cardiovascular physiology can influence treatment outcomes, making personalized approaches increasingly important.
Cardiovascular Bio Digital Twins could eventually help address this challenge by providing patient-specific models that allow researchers to investigate treatment responses in a virtual environment.
For example, a physician could potentially compare how a patient’s cardiovascular system responds to two different medications or evaluate multiple dosage levels using the individual’s digital twin.
Such simulations could help generate additional information about possible treatment effects before decisions are made in clinical practice.
The technology could also support research into complex cardiovascular conditions by allowing scientists to examine physiological changes under different scenarios.
However, these applications remain potential future uses. Additional research, clinical validation, and regulatory considerations will be necessary before cardiovascular digital twins can become routine tools in patient care.
Improving the Speed of Cardiovascular Clinical Modeling
The speed at which a personalized digital twin can be created is a critical consideration for its eventual clinical adoption.
In some healthcare situations, particularly acute cardiovascular conditions such as heart failure, timely assessment and treatment decisions can be important.
A digital twin that requires extensive computational processing may be difficult to integrate into clinical workflows where physicians need information quickly.
NTT Research’s precomputed virtual population approach is designed to address this limitation by shifting much of the computational work to an earlier stage.
Instead of generating every possible cardiovascular configuration during the personalization process, the research team has already created a library of simulated physiological profiles.
This reusable infrastructure could allow individual models to be generated more quickly when patient measurements become available.
The approach may also support larger research studies, virtual clinical trials, and future clinical decision-support technologies.
For healthcare technology companies and research organizations, the concept of reusable digital infrastructure could have broader implications.
A virtual population developed once could potentially support the generation of numerous individual cardiovascular models, reducing the need to repeat extensive simulation processes for every new patient.
This could improve the scalability of cardiovascular digital twin research and create opportunities for evaluating treatment strategies across diverse virtual populations.
Supporting Precision Medicine and Virtual Clinical Trials
Precision medicine aims to tailor healthcare decisions and treatments according to the characteristics of individual patients.
In cardiovascular medicine, this approach is particularly relevant because patients with similar diagnoses may have different physiological profiles and treatment responses.
Cardiovascular Bio Digital Twins could support precision medicine through two major areas of application: clinical research and individualized patient care.
In research settings, digital twins could help scientists investigate how different patient profiles respond to specific therapies. Researchers could simulate multiple treatment scenarios, compare medication strategies, and examine potential outcomes before proceeding with more extensive clinical investigations.
These models may also contribute to virtual clinical trials, where computer simulations are used to explore treatment effects across modeled patient populations.
Virtual trials could eventually help researchers identify promising therapeutic strategies, improve study design, and generate additional evidence to support conventional clinical research.
In clinical practice, personalized cardiovascular models could potentially provide physicians with an additional tool for evaluating treatment options.
For example, a physician might use a patient’s digital twin to simulate the possible cardiovascular effects of different medications or dosage levels. The resulting information could help support discussions about treatment selection.
Digital twins may also offer opportunities to estimate certain internal cardiovascular measurements that are difficult to obtain through routine noninvasive assessments.
Nevertheless, these applications will require extensive validation to establish their reliability and clinical usefulness. Digital twin outputs would need to be evaluated alongside established diagnostic procedures and clinical judgment.
Future Research in Heart Failure and Other Cardiovascular Conditions
The development of a virtual population representing healthy cardiovascular systems provides an important starting point for future research into disease-specific models.
Understanding normal cardiovascular function is essential for identifying how disease changes the behavior of the heart and blood vessels.
By establishing a healthy physiological baseline, researchers can investigate how conditions such as heart failure affect cardiovascular measurements and the relationships between different physiological parameters.
NTT Research plans to explore additional applications of its technology, including virtual clinical trials, expanded drug-effect libraries, and methods that could reduce reliance on invasive right-heart catheterization.
Right-heart catheterization is an invasive procedure used to measure pressures and other hemodynamic parameters within the right side of the heart and pulmonary circulation. Although it provides important clinical information, the procedure requires specialized equipment and medical expertise.
The research team’s longer-term objective includes investigating whether cardiovascular digital twins could help estimate certain physiological measurements that might otherwise require invasive procedures.
Such developments could potentially support more efficient cardiovascular assessment and reduce the need for some invasive diagnostic interventions, subject to appropriate clinical validation.
The company also aims to develop digital twins that can be updated as a patient’s health changes over time.
A continuously updated model could eventually help researchers and physicians monitor changes in cardiovascular function, identify potential deterioration, and evaluate opportunities for earlier intervention.
NTT Research’s Vision for Predictive Healthcare
Joe Alexander, Jr., M.D., Ph.D., F.A.C.C., Director of the NTT Research Biomedical Engineering Innovation Lab, emphasized the importance of moving healthcare toward prevention and prediction.
He stated that the next frontier of medicine involves preventing disease rather than responding only after health problems develop. In his view, creating a cardiovascular Bio Digital Twin for every patient could help shift healthcare from reactive treatment toward more predictive approaches.
This vision reflects the broader potential of personalized digital modeling technologies.
Rather than relying exclusively on measurements collected during individual medical appointments, future digital twins could potentially provide continuously updated representations of a patient’s physiological condition.
Such systems may help identify changes in health earlier and provide additional information for preventive healthcare strategies.
Beyond cardiovascular applications, the underlying modeling approach may eventually be extended to other organs, including the lungs and kidneys.
Developing digital representations of multiple organ systems could provide opportunities to investigate interactions between different physiological processes and support more comprehensive approaches to personalized medicine.
Research Team and Publication
The research findings have been published in Frontiers in Physiology under the title “Accelerating Cardiovascular Digital Twin Creation Using Precomputed Virtual Populations as Bayesian Priors.”
The paper’s authors include:
- Iris L. Shelly – Lead and corresponding author
- Jon N. Peterson
- Giulia Cornali
- Jennifer Rohrs
- Patrick E. Gelbach
- Joe Alexander, Jr.
All authors are affiliated with the NTT Research Biomedical Engineering Innovation Lab, except Jennifer Rohrs, who is affiliated with Embody Biosciences.
The publication presents the scientific foundation of the precomputed virtual population approach and its application to accelerating cardiovascular digital twin creation.
Advancing the Future of Personalized Cardiovascular Care
NTT Research’s latest development represents progress toward making personalized cardiovascular digital twins more scalable and practical for future healthcare applications.
By combining physiological modeling, precomputed virtual populations, and Bayesian methods, the Biomedical Engineering Innovation Lab has developed an approach intended to reduce the computational burden associated with creating patient-specific cardiovascular models.
The research could support future investigations into personalized treatment selection, virtual clinical trials, cardiovascular disease monitoring, and predictive healthcare.
Although additional research and clinical validation will be required before the technology can be routinely used in patient care, the approach establishes a foundation for exploring how digital representations of individual organs and physiological systems could contribute to precision medicine.
As NTT Research continues developing its cardiovascular modeling capabilities, its longer-term vision is to support healthcare approaches that identify potential problems earlier, improve understanding of individual patient physiology, and help physicians make more informed treatment decisions.
The company’s work also highlights the growing role of biomedical engineering and computational modeling in transforming how cardiovascular diseases may be studied, monitored, and managed in the future.
About the NTT Research Biomedical Engineering Innovation (MEI) Lab
The NTT Research Biomedical Engineering Innovation (MEI) Lab exists to bring cardiovascular care and precision medicine to patients anywhere, from a war zone to the back of an ambulance, even before a crisis occurs. Through research in cardiovascular Bio Digital Twins, cardiovascular care, and advanced precision medicine, the Lab develops technologies that predict disease, personalize care, and improve outcomes.
The NTT Research Biomedical Engineering Innovation (MEI) Lab is inventing a future of healthcare that moves from treatment to prevention by identifying disease earlier and delivering more precise interventions.
About NTT Research
NTT Research is the Silicon Valley research arm of NTT, one of the world’s largest technology and business solutions providers. Founded in 2019, NTT Research advances foundational science and accelerates its real-world impact across NTT’s global ecosystem. As part of NTT Group, a global enterprise with more than $90 billion in annual revenue, serving more than 75% of the Fortune Global 100 and investing approximately $3 billion annually in research and development, NTT Research is positioned to move deep science from the laboratory toward global-scale application.