It’s no secret that most biomedical firms today use modeling to make research and development decisions. What remains to be seen is how to take modeling, within companies and among regulatory agencies, to the next level.
It’s no secret that most biomedical firms today use modeling to make R&D decisions. What remains to be seen is how to take modeling, within companies and among regulatory agencies, to the next level.
“We would like to eliminate the need for some regulatory testing, but how do we convince reg body to use this information?” said Clifford Warner, a senior materials science engineer at W.L. Gore & Associates, at the recent AABME CONNECT conference, where many sessions and conversation focused on building trust in modeling. “What we do is complex. The language we use confuses people outside of the field. It shuts people down because of sheer complexity of it.”
Chris Mullin, director of product development strategy at NAMSA, a contract medical research organization, answered: “We build trust by building useful models,” he said. “When code and data are transparent, we don’t need personal trust as much. In god we trust; all others bring data.”
Taylor also noted the need to break down barriers within companies. He urged modelers to build credibility by collaborating with respected engineers within their organization on building their models.
Representatives from other regulatory agencies spoke at CONNECT, held last month in Minneapolis, MN, about how they use models in the regulatory process. The conference was held in conjunction with ASME’s Verification and Validation (V&V) Symposium.
When it comes to modeling, industry needs standards on how to build, verify, and validate models, said David Moorcroft, a research mechanical engineer at the Federal Aviation Administration. But he does believe the industry is on the verge of change. The biggest challenge grows from the fact that FAA certification programs are run by independent companies. The value of their models is that they can use them again, but they have yet to work out a business model that allocates the cost of the model between the customer who needs the models now and customers that will use it in the future.
“To trust simulations, you need to convince yourself that their output is on par with results of real world data,” said Joshua Kaiser, a reactor engineer with the Nuclear Regulatory Commission. “You need to acknowledge that you can and will make mistakes in the simulation. A good method to check those mistakes will fix a bad code, but having no method to run those checks will ruin a good one.”
The final panel looked at the future of modeling and simulation.
Jeff Bischoff, extremities research manager at Zimmer Bionet, argued at a panel on the future of modeling and simulation that the barrier to progress is not technology. “Have lots of data and processing power,” he said. “The critical issue is making sure we put conscious thought power behind this.”
James Thompson, director of industrial strategy for medical devices at Siemens Product Lifecycle Management Software, pointed to the enormous range of disease conditions. “The big challenge is, how do solve problems that are so big and so wide,” he said. “Everything is a little bit different. How do we organize our efforts to make more holistic progress?”
“People need a rallying point, a success story they can build upon,” added F. Scott Gayzik, an associate professor at Wake Forest School of Medicine. “We need something to choose as a foothold and build on that.”
The panel also discussed what lessons device makers might learn from other industries. Gayzik pointed to the automotive industry. “The reason they got into crash dummies was that it was easier,” he said. “They simplified headforms and then others used them in drop tests in sports to test helmets.
Thompson reemphasized the point. The reason automotive and aerospace models are so effective has little to do with technology.
“They take it as a given that you can come up with right model if you can correlate it to physical test results,” he said. “But in those industries, the big investment they made was not in advancing FEA modeling technology, but in doing a lot of correlation testing so they are really confident their models are accurate.”