Harnessing the Potential of Digital Twins

Combining physical systems with their digital counterparts could enhance decision-making, though concerns about reliability and trustworthiness remain.

Imagine a scenario where a doctor is evaluating test results for two cancer patients. One patient is a 62-year-old woman waiting in the exam room, while the other is her virtual counterpart — a series of simulations and models replicating her and the tumor. This system, known as a digital twin, integrates real-time data and simulations to mirror the patient’s condition and predict treatment outcomes.

Medical tests and imaging results continuously update the virtual model, allowing simulations to predict how the patient might respond to various treatments. This approach can minimize invasive procedures for patients, tailor personalized care plans, improve outcomes, and reduce healthcare costs.

Recent advancements in digital twin technology — which create virtual representations that replicate the structure, context, and behavior of physical systems — have brought this vision closer to reality. Unlike traditional simulations, digital twins offer continuous feedback between virtual and physical components, facilitating real-time updates and interactions.

The demand for digital twins is expanding across various fields. For instance, a digital twin of a city’s transportation network could reduce traffic congestion, forecast the impact of new bus routes, and guide infrastructure investments. Similarly, a digital twin of a coastal area could help emergency planners and residents anticipate climate change impacts and enhance disaster preparedness and recovery efforts.

However, for these applications to become widespread, decision-makers need assurance that digital twins are reliable and trustworthy, especially for critical and safety-related uses. Establishing credibility and trust is crucial, according to a 2023 report by the National Academies titled “Foundational Research Gaps and Future Directions for Digital Twins.”

“Digital twins have immense potential across various fields, including engineering, natural sciences, and medicine,” said Karen Willcox, director of the Oden Institute for Computational Engineering and Sciences at the University of Texas at Austin and chair of the committee that authored the report. “Addressing significant research questions and ensuring responsible development is key to maintaining trust in these technologies.”

Assessing Digital Twin Reliability

Digital twins rely on dynamic updates from real-world data, such as sensors, clinical assessments, or remote sensing. These updates inform simulations that can drive changes in the physical system, like adjusting a patient’s medication dosage or activating a weather drone sensor.

In some instances, humans will make decisions based on digital twin insights, such as a doctor choosing a treatment plan. In other cases, systems might be partially or fully automated, like a digital twin of an aircraft adjusting sensors to enhance data quality.

According to the 2023 report, assessing the reliability of digital twins involves verification, validation, and uncertainty quantification (VVUQ) processes. These methods evaluate the accuracy of digital twins and the quality of their predictions.

Developed over the past several years, VVUQ processes support increasingly complex simulations, including those involving machine learning and artificial intelligence. However, digital twins present new challenges for VVUQ due to their dynamic nature. “We need new methods that can adapt to changes in models, data, and decision contexts,” Willcox noted.

The report also highlights the lack of standard procedures for reporting VVUQ results, which complicates decision-makers’ ability to trust modeling outputs. The report advocates for embedding VVUQ deeply in digital twin technologies from design to deployment.

An upcoming National Academies symposium, scheduled for June 17, will delve into these issues. Titled “Assessing the Reliability of Complex, Dynamic Modeling and Simulation,” the event will explore how to measure and evaluate the reliability of digital twins and other complex modeling systems, as well as the limitations of current VVUQ methods.

“The systems targeted by digital twins are often complex, with models containing numerous uncertainties,” said Omar Ghattas, Fletcher Stuckey Pratt Chair in Engineering and director of the OPTIMUS Center at the Oden Institute for Computational Engineering and Sciences, and a symposium moderator. “Given their critical importance to society, it is essential to rigorously account for uncertainties across the digital twin lifecycle, from data assimilation to model-predictive decision-making.”

By addressing these challenges and ensuring robust VVUQ processes, digital twins can fulfill their promise of enhancing decision-making and improving outcomes across various sectors.