The Expanding Role of Digital Twins in Medical Design & Manufacturing

Digital twins (DTs)—virtual duplicates of objects, products, processes, and even entire corporate ecosystems that physically exist and evolve using data from their operational surroundings—continue to transform manufacturing through operational optimization, predictive maintenance, and reduced risk. DTs span an object’s lifecycle, are updated with real-time data, and rely on simulation, machine learning (ML), and reasoning to help make decisions. 

A common misconception is that the goal of digital twins is to recreate reality in every detail. “In practice, however, the most effective digital twins are designed with a much more focused objective: to capture the parts of a system that matter for the decision an organization is trying to make,” said David Andersen, regional director for Oakbrook, Ill.-based The AnyLogic Company, a developer of simulation modeling software for business applications. “A model designed to improve manufacturing performance may look very different from one built to evaluate supply chain resilience, production capacity, or product launch strategy.” 

At their core, digital twins are not simply digital replicas. They are purpose-built decision-support environments that help organizations better understand complex systems before making changes in the real world. Their real value is in creating a controlled environment where organizations can test ideas, evaluate tradeoffs, and understand the consequences. “Organizations are increasingly building digital twins that evolve alongside the products, manufacturing systems, and business questions they are intended to support,” said Andersen.

For example, in product development, DTs help engineers evaluate design alternatives, predict performance, identify quality issues, and improve manufacturability. In manufacturing, digital twins can optimize production scheduling, capacity planning, equipment utilization, and operational performance while reducing scrap, downtime, and time to market.

“Increasingly, leading manufacturers are integrating digital twins into sales inventory operations planning [SIOP], allowing leadership teams to evaluate demand, capacity, inventory, supplier constraints, and financial tradeoffs before making critical business decisions,” said Lisa Anderson, president of LMA Consulting Group, a Claremont, Calif.-based firm that specializes in manufacturing strategy and end-to-end supply chain solutions. “When combined with artificial intelligence, Internet of Things, and advanced planning systems, DTs are evolving from engineering tools into strategic business capabilities that improve predictability, resiliency, customer service, and overall operational performance.”

Digital Twin Trends

“Digital twins are transforming medical device engineering by shifting development from a reactive build–test–fix model to a more proactive predict–optimize–validate approach,” said Mohan Ramadoss, director of global innovation and development for Phillips Medisize, a Molex company in Hudson, Wis., that provides design, engineering, and manufacturing services to the medical device, pharmaceutical, and in-vitro diagnostics industries. “Rather than a standalone simulation, DTs act as a system-level predictive framework, integrating multi-physics models, system behavior, physical prototyping, and real-world data to evaluate performance under realistic conditions.”

In the early design phase, for example, a digital twin can enable rapid design-space exploration and sensitivity analysis, linking outputs from digital and physical prototypes—such as dose delivery, force, pressure, and flow—to key design variables. This can support structured design of experiments (DOE), allowing engineers to understand patterns, interactions, and dominant influencing variables—driving scientifically guided design decisions rather than trial-and-error approaches, particularly when adapting device platforms for different drug molecules.

“Increasingly, digital twins are extending into the manufacturing sector, linking design intent with process variation—such as molding, assembly tolerances, and material behavior—to predict final device performance,” said Ramadoss. “This can support tolerance allocation, manufacturability assessment, and early failure-risk identification, ensuring designs are robust before major tooling investments.”

The biggest shift that Anderson has seen in medical device manufacturing is that DTs are no longer viewed primarily as engineering or simulation tools. Instead, they are becoming enterprise decision-support platforms that connect engineering, manufacturing, supply chain, and business planning. “They are moving from product twins to enterprise twins,” said Anderson. “Companies are using them to create digital representations of production lines, factories, and even end-to-end supply chains to optimize performance across the business.” 

Digital twins, when integrated with artificial intelligence (AI), can identify constraints, quickly evaluate thousands of scenarios, predict outcomes, and recommend actions. “AI and digital twins are often discussed together, but they solve different problems,” said Andersen. “AI is exceptionally good at identifying patterns, forecasting demand, and detecting anomalies. Simulation helps organizations understand how decisions will affect complex systems before those decisions are implemented.”

However, to get the best results, DTs must be applied with discipline. “They are bounded predictive models, not exact replicas of reality,” said Ramadoss. 

Their value lies in enabling credibility-driven modeling, where a clearly defined context of use, validation evidence, and quantified uncertainty guide decision-making. In this framework, he noted, digital twins do not replace testing—”they complement physical prototyping by making testing smarter, more focused, and more efficient.”

Humans in the Loop

Digital twins in medical device development are rapidly evolving from open-loop, component-level simulation models into closed-loop, system-level predictive frameworks. Traditionally, digital twins were used to predict outputs—such as dose delivery, pressure, or flow—based on controlled design inputs such as geometry, material properties, and actuation profiles. “However, this approach assumes a fixed environment and does not fully capture real-world variability,” said Vasanthan Mani, senior manager of product development for Phillips Medisize.

A key advancement is the emergence of human-in-the-loop digital twins, particularly for complex drug delivery systems such as on-body injectors and high-viscosity injectors. In these systems, dose delivery performance is no longer governed solely by pump mechanics or fluid dynamics, but by nonlinear interactions between device, fluid rheology, and tissue mechanics. Engineers are now incorporating simplified anatomical models that include tissue compliance, back pressure, and fluid–structure interaction into the simulation loop. “This can enable closed-loop prediction, where device response dynamically adapts to physiological boundary conditions, providing a more realistic estimate of injection time, flow variability, and end-of-dose accuracy,” said Mani.

From product launches to M&A and manufacturing trends, the MPO Source delivers the medtech news you need twice a week. Subscribe here!

Medical device manufacturers (MDMs) use DTs to create connected, real-time models that integrate product development, manufacturing, quality, supply chain, and business planning into a single decision environment. “For MDMs, that means combining engineering data with production schedules, supplier performance, inventory, quality metrics, and customer demand to evaluate multiple scenarios before decisions are implemented,” said Anderson. “Instead of reacting to disruptions, companies can anticipate the operational, financial, and customer impacts of alternative decisions and choose the best path forward.”

MDMs are intensely focused on achieving early confidence in product performance. In essence, digital twins are hybrid predictive systems that help MDMs do just that—a physics-based core augmented, where useful, by AI/ML surrogates. “They will be disruptive to the industry not because they replace existing processes, but because they enable a shift from reactive, test-driven development to predictive, system-driven engineering—compressing timelines, reducing cost, and increasing confidence across the entire product lifecycle,” said Vishwas V. Bhat, senior manager of global engineering services for Phillips Medisize.

Regulations and Standards

Navigating the regulatory landscape for DT technology can be challenging, with stringent requirements for validation, performance, and safety. On the upside, however, the digital twin framework can also reduce the time and cost associated with clinical trials and regulatory approvals “by simulating regulatory testing environments and reviews,” reported Quality Magazine. “MDMs can use this framework to simulate the performance of a new device under various conditions, ensuring compliance with safety standards such as IEC 60601 before the device undergoes physical testing and regulatory review.”¹

Establishing industry-wide standards for DTs is also important for ensuring interoperability across various systems and applications. The ISO/TC 184 technical committee is leading efforts to standardize digital twins—for example, the ISO 23247 standards were developed to provide guidance on digital twin frameworks.

“Regulatory acceptance of digital twins is starting to evolve around the principle of model credibility, rather than model complexity,” said Ramadoss. “Digital twins are increasingly used to support submissions, provided they are aligned with a clearly defined context of use and supported by appropriate validation data from the real system and uncertainty quantification.”

Rather than replacing physical testing, digital twins are best positioned as part of a hybrid validation strategy, where simulation complements targeted testing to build traceable, evidence-based justification. This approach aligns with FDA guidance on computational modeling, which emphasizes verification, validation, and credibility assessment.

A hybrid DT now spans two frameworks: the physics core falls under the 2023 final CM&S guidance/ASME V&V 40, while AI/ML components fall under the FDA’s separate January 2025 draft on AI credibility. “In practicality,” said Ramadoss, “where the AI sits matters.”

The FDA’s Medical Device Development Tools (MDDT) program lets a model be qualified once and reused across submissions and early Q-submission engagement, which establishes the acceptable computational approach up front. As a result, digital twins are enabling a shift toward designing for regulatory acceptance, improving both efficiency and confidence in approval pathways.

Personalized Medicine

MDMs use digital twin technologies to widen the range of applications for personalized medicine by integrating an individual patient’s own health data. DT models help customize therapeutic strategies by simulating individual physiological responses, potentially reducing trial and error in treatment selection. 

The Association for Advancing Automation states that “one of the most compelling applications of digital twins in healthcare is in the realm of individualized patient modeling, which integrates diverse health data to build a continuously updated model of an individual’s physiology. In principle, such a twin could help clinicians understand disease progression, compare a patient’s trajectory against population data, and design more tailored treatment plans.”²

Digital replicas of organs (especially the heart) are used in clinical and research settings to simulate responses to therapy. For example, models built from magnetic resonance imaging and electrocardiogram measurements can simulate cardiac mechanical and electrical behavior to visualize how treatment choices might impact outcomes, before a real intervention is attempted, reducing risk and improving planning.²

In another example, GE HealthCare has partnered with Mayo Clinic to launch a digital twin project for personalizing radiation cancer therapy that combines imaging, AI, and patient monitoring. The project—called GEMINI-RT—will utilize DTs to improve the delivery of radiation doses to tumors, avoiding healthy organs. The first step will connect treatment to initial diagnostic scans and create digital models of patients. GEMINI-RT is grounded in the concept of “twinning the patient, personalizing the beam.” Bryan Traughber, Mayo Clinic’s vice chair for innovation in radiation oncology, explained, “Our goal is to model individual patient journeys with precision, enabling radiation therapy treatments that are truly tailored to each patient.”³

The High Energy Photonics Center

The High Energy Photonics Center (HEP), a Siemens Healthineers facility in Forchheim, Germany, that opened last year, was conceived as a fully integrated environment for developing, producing, and continuously improving highly complex medical technology components. 

“From the outset, our objective was not to simply build a new factory, but rather to establish a seamless digital thread connecting product development, industrialization, manufacturing, logistics, and operations,” said Dr. Markus Kaupper, head of digitalization for HEP and senior engineering and supply chain leader for Siemens.

Digital twins played a key role throughout the design process, noted Dr. Kaupper. “We created virtual representations of products, production systems, material flows, automation concepts, and factory operations—long before the physical facility was built.” This allowed design teams to simulate manufacturing scenarios, validate production concepts, optimize layouts, and evaluate automation strategies before implementation.

“As our twinning capabilities increased, we gained deeper insights into our processes and product performance—unlocking new potentials and at a speed that was not possible before,” said Dr. Kaupper. “While data science projects took several months in the past, we are now able to introduce new machine learning applications within weeks or even days—a game-changer for us.”

A product example is X-ray tubes for imaging systems, which require highly sophisticated physical processes, stringent quality requirements, and close collaboration across engineering, production, and quality functions. Digital twins help connect product engineering and manufacturing execution. Product characteristics, design parameters, process specifications, and quality data are digitally linked across the entire product lifecycle. The digital twins enabled engineers to simulate concepts and assess manufacturability long before physical prototypes were built. During industrialization, production processes and equipment can be validated in a virtual environment. The pilot rollout was quite successful—for example, reaction times were projected to improve by 20% to 25% on the production floor.

“Once production begins, process data and manufacturing feedback will flow back into engineering, creating a continuous feedback loop that drives quality improvements and innovation,” said Dr. Kaupper.

Other Collaborations

GE HealthCare and NVIDIA have collaborated to develop “Isaac for Healthcare”—a platform purpose-built for developing healthcare robots.⁴ Built on NVIDIA’s three-computer framework for AI, Isaac for Healthcare is a platform designed to help developers build, simulate, and deploy AI-powered medical robots. It uses digital twin environments and synthetic data to allow systems to practice surgeries, scans, and treatments. This allows developers to train and validate autonomous imaging capabilities (for example, automated patient positioning and X-ray or ultrasound quality) in a virtual, physics-based environment before becoming part of clinical hardware. 

By combining the power of digital twins and AI, Isaac for Healthcare makes it possible for medical device designers and engineers to: 

• Explore digital prototyping of next-generation healthcare robotic systems, sensors, and instruments
• Train AI models with real and synthetic data generated by ‌high-fidelity simulation environments
• Evaluate AI models in a digital twin environment with hardware-in-the-loop (HIL)
• Collect data for training robotic policies through imitation learning by enabling extended reality and/or haptics-enabled teleoperation of robotic systems in digital twins 
• Train robotic policies for augmented dexterity (for example, robot-assisted surgery) and use graphics processing unit parallelization to train reinforcement and imitation learning algorithms
• Continuously test robotic systems through HIL digital twin systems
• Create deployment applications to bridge simulation and deployment on a physical surgical robot

The Isaac for Healthcare platform also features a variety of specific tools and capabilities for high-level R&D needs. For example, developers can convert computed tomography or magnetic resonance imaging scans into realistic 3D patient models, eliminating the common problem of limited medical training data. The platform’s high-fidelity sensor simulation generates synthetic ultrasound and camera data so AI can learn across a huge range of possible scenarios. 

Moving Forward 

“The biggest advances will not come from creating more sophisticated digital twins,” said Anderson. “Instead, they’ll come from making them better decision-support tools—for example, integrating them into SIOP and executive decision-making. Instead of simply monitoring operations, manufacturers will use digital twins to evaluate how changes in customer demand, capacity, supplier performance, inventory, quality, and financial objectives affect business outcomes before decisions are implemented. Companies that leverage digital twins this way will make faster, better-informed decisions, improve predictability, and gain a significant competitive advantage.”

The next generation of digital twins will be enabled by the convergence of reduced-order modeling, AI-accelerated simulation, and closed-loop system integration. Reduced-order and 1D system models can capture essential behavior while allowing rapid evaluation across large design spaces, supporting fast parametric studies and virtual DOE.

“At the same time, AI/ML-based surrogate models are emerging as high-speed predictors within validated boundaries, significantly accelerating optimization workflows,” said Mani. “The industry is adopting a hybrid approach, where physics ensures accuracy and interpretability, while AI improves scalability and speed—transforming DTs into efficient, decision-grade platforms.”

A particularly important advancement is the shift toward closed-loop, human-in-the-loop digital twins, where physiological factors such as tissue compliance and back pressure are incorporated into the system. This can enable prediction under real-world patient variability, which is critical for complex delivery systems.

In parallel, improvements in multi-physics co-simulation and growing availability of material and reliability data are enhancing predictive fidelity in previously difficult areas. “Combined with integration of field and manufacturing data, digital twins are evolving toward continuous learning systems that improve over time,” Mani added.

“We are also seeing an increased emphasis on resilience,” said Andersen. Recent disruptions across global supply chains have highlighted how interconnected medical device manufacturing has become. “Organizations increasingly want to understand not only how to optimize operations under ideal conditions, but also how systems behave when suppliers fail, demand changes unexpectedly, equipment goes offline, or production priorities shift,” said Andersen.

Driven by automation, AI, and DT technologies, Manufacturing 4.0 is emerging as a new paradigm applied to regulated healthcare environments such as pharmaceutical plants, medical device factories, laboratory operations, and connected care ecosystems. Unlike traditional manufacturing models that rely on periodic reporting and largely decentralized systems, Manufacturing 4.0 relies on real-time visibility, machine connectivity, simulation, adaptive automation, and analytics-driven decision making, forming a “digital twin value chain,” said Deepak Prakash, vice president of healthcare at Identiv, a Santa Ana, Calif.-based provider of RFID- and BLE-enabled IoT solutions that create digital identities for physical products across industries.

In healthcare manufacturing, this matters because performance is measured not only by efficiency but also product quality, patient safety, validation rigor, and regulatory compliance. “Key enablers include cloud and edge computing, robotics, advanced analytics, and artificial intelligence,” he added. “Together, these and other capabilities create the conditions for digital twins to move from theoretical models to real-world deployment examples at scale.”

References

1. tinyurl.com/mpo260751
2. tinyurl.com/mpo260752
3. tinyurl.com/mpo260753
4. tinyurl.com/mpo260754

Mark Crawford is a full-time freelance business and marketing/communications writer based in Corrales, N.M. His clients range from startups to global manufacturing leaders. He has written for MPO and ODT magazines for more than 15 years and is the author of five books.