Manufacturing technology
performance falls outside its validated or acceptable range, the consequences may include increased risk to product quality, manufacturing operations or, where relevant, patients. To combat the potential impacts of model drift, Schoen stresses the importance of shifting away from one-time validation to a continuous approach, which Kaminski refers to as “validation as a monitored state, and not a milestone”.
This may require documenting the model’s intended use, inputs, outputs, limitations and relevant decision process, alongside risk-based monitoring for performance or data changes. It may also involve tracking the lineage, acceptance criteria and original source of the data used to train the AI. While Sainath agrees with Kaminski and Schoen, noting that validation is “extremely critical” for models using unsupervised learning, she qualifies this by noting that Siemens Healthineers mostly employs supervised- learning models trained on curated, labelled datasets. She says the company does not currently treat these systems as autonomous decision-makers.
Sometimes, less is more As AI continues to advance rapidly, manufacturers have access to an expanding range of models they can embed into their workflows, with varying levels of complexity. On responsible AI implementation, Kaminski notes that the focus should not necessarily
be on adopting the newest or most advanced technologies, but on implementing specific, vertical tools best suited to their intended purpose. By choosing the right AI for the job, Kaminski says manufacturers may reduce the computing needed to run a model compared with a larger general-purpose alternative. However, the resulting energy and water use depends on the model, hardware, workload, data- centre conditions and deployment pattern. Sainath echoes Kaminski’s point, noting that Siemens Healthineers opts to implement models based not on compute power or complexity, but on their real-world applications and benefits within a narrow context. This approach also aligns with the company’s focus on sustainability in its manufacturing processes, as models with lower operational resource requirements may carry a smaller environmental burden, subject to their full life-cycle impacts. The company says reducing environmental impact is a priority, alongside wider policy and industry efforts to improve the sustainability of manufacturing and resource use.
When considering AI implementation, Schoen believes advice from a relevant expert can help organisations assess which models are appropriate for a specific context. Ultimately, fit-for-purpose AI, supported by appropriate validation, monitoring and human oversight, will still be important to the future use of AI in medical device manufacturing. ●
50
U.S. Medical Device Electronic Contract Manufacturing
TRICOR offers software, electronic and electro-mechanical system development from concept to full production.
Design Develop Manufacture
Come visit us at MD&M Midwest, Booth #3347 ISO 9001:2015
ISO 13485:2016 1-847-742-5542
www.tricor-systems.com 40
www.medicaldevice-developments.com
CMMC Level 2
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100 |
Page 101 |
Page 102
