This article was originally written for and published by .
Not surprising, due to its conservative business model, the medical device industry has been slow to embrace artificial intelligence (AI), according to a .
Only 24% of 200 respondents reported active use of AI for product operations.
鈥淔or business efficiency, medical device companies are behind their peers and are not taking advantage of some of the modern technology,鈥 said Ross Meyercord, CEO of Propel Software, which commissioned the survey conducted by Talker Research in May.
, based in Redwood City, CA, collaborates with businesses to help transform the way they design, manufacture, and deliver products to customers.
Three other industries were also polled: consumer goods, high tech, and industrial equipment. All three of these industries had a higher percentage of AI use in their product processes to increase productivity and efficiency, with high tech having the greatest rate of AI adoption in their product operations (51%).
One of the benefits of AI for the medical device industry is allowing products to become smarter, safer, and more efficient 鈥渋n doing the job that they鈥檙e designed to do,鈥 Meyercord told MD+DI. Another advantage is AI鈥檚 ability to interpret data to better diagnose medical conditions earlier in their disease progression.
A third area, and the one Propel is most focused on, 鈥渋s leveraging AI to make the design and quality processes more efficient in the manufacture of medical device products,鈥 Meyercord said.
The survey found that medical device companies actively using AI have achieved increased productivity (44%), a greater competitive advantage (35%), reduced expenses (31%), technology consolidation (25%), resource reallocation (25%), and lower headcount (23%).
Conversely, for those companies not leveraging AI across product development, the biggest challenge was team coordination (33%), followed by no formal process for gathering customer or market feedback to inform product decisions (25%) and acquiring product data and specifications (23%).
鈥淎I is good at finding patterns in the data, but if your data itself has no patterns because it is incomplete or erroneous, then AI鈥檚 power will be significantly limited.鈥
One of the hurdles when considering AI is the need to contract with an AI third party to validate AI solutions for quality processes and design tools, for example.
鈥淏ut there are some providers, like Propel, where the is built into the software package itself,鈥 Meyercord said. 鈥淲e offer a validated solution set that includes AI, while still allowing the customer to meet all necessary regulatory requirements to operate effectively.鈥
However, the operational challenges of AI can be daunting. What is the status of your data? 鈥淭he old adage of garbage in equals garbage out definitely applies to the world of AI,鈥 Meyercord said. 鈥淎I is good at finding patterns in the data, but if your data itself has no patterns because it is incomplete or erroneous, then AI鈥檚 power will be significantly limited.鈥
Medical device companies with the most success in incorporating AI review their sources of data and purge old and 鈥渄irty鈥 data from the system. 鈥淭he focus should be: How can we enrich the data and provide more value to it, and then apply AI to that data set,鈥 Meyercord said.
Rethinking workflows and how the workflow conforms in the context of AI is also key. 鈥淚f you鈥檙e leveraging AI to perform some tasks and the old workflow where the human did steps 1 through 10 now has AI doing steps 2, 5, and 8, your workflow may no longer make sense,鈥 Meyercord said. 鈥淵ou may want to redesign it so that the AI steps are done contiguously in that process. A human may start and end the process, with AI for the steps in between.鈥
Incorporating AI in small, incremental steps is preferred, according to Meyercord. 鈥淟earn from your experience and adjust for the next round,鈥 he said. 鈥淭rying to treat AI as a large transformation project out of the gate can be very difficult.鈥







.png)
%20(1).png)



