Scaling In Silico Clinical Trials with Cloud Computing

Medical device innovation depends on generating evidence that new designs are safe and effective. Other than pre-clinical testing, evidence traditionally comes from clinical trials that can take years to complete and often involve relatively small patient populations. For orthopedic implants, this creates a challenge: understanding how a device will perform across the full spectrum of patient anatomies, bone qualities, and surgical techniques can be difficult using conventional studies alone.

To help address this challenge, Zimmer Biomet collaborated with Ansys, part of Synopsys, Amazon Web Services (AWS), Invisibl Cloud, and Dassault Systèmes to dramatically scale an in silico clinical trial (ISCT) for a shoulder replacement implant. By moving an existing, highly complex simulation workflow to the cloud, the team transformed a process that previously took months into one capable of delivering results in less than a day while analyzing a population 45 times larger than before. This represents an approximately 100-fold increase in simulation throughput.

“Simulations enable direct evaluation of multiple interventions on the same virtual patient, providing insights that cannot be achieved in a traditional clinical trial,” says Jeff Bischoff, Senior Director of Biomechanics Research at Zimmer Biomet.

Why In Silico Clinical Trials Matter

In silico clinical trials use computer simulations rather than human participants to evaluate medical devices. Engineers create large populations of realistic virtual patients and simulate surgical procedures and outcomes using advanced computational models.

“The approach offers several advantages,” says Philippe Favre, Associate Director of Computational Biomechanics at Zimmer Biomet. “Virtual populations can be intentionally designed to represent a broad range of patient characteristics, including uncommon anatomies and edge cases that may be difficult to capture in traditional studies. Simulations can also be completed in weeks or months rather than years, helping development teams evaluate designs more quickly and accelerate innovation.”

“Historically, however, the scale of these studies has been limited by available computing infrastructure and software licensing constraints,” says Sandeep Sovani, Worldwide Go-To-Market Lead for Cloud-Based Modeling & Simulation at AWS. “Even relatively modest trials involving a few thousand simulations can require months of processing time, restricting their use during active product development. This is where a cloud provider like AWS brings the missing piece to this important innovative process.”

Starting with a Proven Workflow

The project built on a previously completed Zimmer Biomet shoulder arthroplasty study (by Maquer et al. 2026). That original trial involved roughly 520 virtual surgeries and required approximately three months to run on internal infrastructure. Using an already validated workflow ensured that results produced in the cloud could be compared against established benchmarks while providing a realistic baseline for measuring performance improvements.

The team's objective was ambitious: increase the size of the original virtual cohort by a factor of 45 while reducing runtime from three months to less than 24 hours.

The Challenge of Scaling

“Simulating a virtual surgical intervention is far more than a single simulation,” says Christine Mueri, Research Manager, Computational Biomechanics at Zimmer Biomet. “It includes a different model for each potential patient harm, with each case involving multiple stages, including virtual implantation, model preparation, meshing, finite element solving, and post-processing. These steps may require different software packages, different operating systems, and different computing resources.”

Scaling this workflow to tens of thousands of simulations introduced several challenges:

• Coordinating thousands of independent workflows simultaneously.
• Ensuring data can move efficiently between processing stages.
• Managing software licensing across multiple commercial applications.
• Handling failures automatically at large scale.
• Maintaining auditability and traceability suitable for regulated environments.

Delivering Cloud-Scale Performance

“At peak operation, the environment was designed to use approximately 35,000 virtual central processing units (vCPU),” says Dhamodharan Pandurangan, CTO at Invisibl Cloud. “Autoscaling capabilities allowed resources to expand and contract based on workload requirements, eliminating the need to permanently maintain infrastructure sized for infrequent large-scale studies.”

The resulting platform demonstrated the ability to execute a virtual trial comprising over 46,000 simulations, representing 23,000 virtual shoulder replacement interventions, in less than 24 hours. Compared with the original three-month, 1,000-simulation study, the new approach enabled a 45-fold larger dataset while dramatically reducing runtime, producing approximately 4,500 times more simulation throughput.

“The platform demonstrated that it can orchestrate and execute tens of thousands of simulation jobs at massive parallel scale while efficiently utilizing a compute fleet of roughly 1,550 workers,” says Gnanaprakash Ramswami VP of Technology at Invisibl Cloud. “It sustained high concurrency and throughput throughout the run, continuously replenished completed work, and successfully drained all 55,846 jobs without leaving residual workload. These results validate the platform's ability to handle demanding HPC-scale batch workloads and provide a strong foundation for further scaling.”

What This Means for Healthcare Innovation

The impact extends beyond faster computation. Large-scale simulation makes it possible to evaluate implant designs across broader patient populations for various surgical protocols and better understand rare scenarios and the input of surgical decision makings and the impact of certain design decisions on the patient’s outcome. Additionally, it generates richer evidence earlier in the development process, all of them contributing to better safety for the patient while dramatically reducing the time to medical innovation. Instead of treating large-scale in silico studies as occasional milestones, engineering teams can soon use them as routine decision-making tools throughout design and optimization.

Ultimately, the project demonstrates how cloud-scale computing can transform computational evidence generation, helping medical device manufacturers evaluate designs faster while maintaining the rigor needed to support innovation and improve patient outcomes.

Learn more about how Ansys can support your healthcare innovations.