Have you robots in place yet?
Martin Dueblin, One One Eleven, asked this question at the beginning of Track A Session 3 “Robotics in Pharma”. Only a few participants use robots in the manufacturing, analyze their data or are on the way to artificial intelligence. He talked about the current status of robots and how they can contribute to the pharmaceutical industry.
Robots facilitate difficult and complex tasks in many industries. Why should this not also apply to the pharmaceutical industry?
You find robots already in
… and tomorrow
Individual companies take the initiative and invest.
Martin Dueblin showed a video of an existing manufacturing plant. The customer started investing in new manufacturing technologies. They explored projects through feasibility studies, prototyping and industrialisation:
PAT, modelling and real-time-release supported these projects.
The presentation focused on robotized batch manufacturing.
Robotics and automation replaced manual operations in batch manufacturing, which is on a similar scale today. The regulatory burden was low, and they minimized open handling steps and improved aseptic processing. The results: Reduced manufacturing costs, smaller footprint, higher flexibility, reproducibility, and quality. It is easy ot introduce elements of process analytical technology (PAT), at-line real time release (RTR) and real time continued process verification (CPV).
Especially aseptic and sterile manufacturing and processing is perfect for robotics and automation. It ensures high precision dosing, liquid transfer, and sterile filtration.
To summarise: Invest in robotics before others do.
Thank you, Martin for this quick insight into the future of drug manufacturing!
In her presentation “Data Management Between R&D and GMP”, Helen Thomas, Janssen Vaccines, reported from an industry study. People waste an average of 2.8 hours a week searching for data, formatting data, and so on. This presentation was part of Track B in Session 4 “Beyond the Quality of Data”.
Further, she found that knowledge gets lost at every handover from R&D to commercial manufacturing. This is a costly problem that can be solved with a product manager who is involved in R&D projects at a very early stage.
Helen Thomas also identified gaps in IT landscapes in the areas of
She recommended integrating and connecting everything to simplify the IT landscape in a company.
In her roadmap to improve digital capabilities, she mentioned three aspects:
Helen Thomas closed her presentation with a look to the people. They have to convince people so they go along with this path and don’t get lost.
It’s a complex problem worth addressing. It increases compliance and ease of use. The system becomes digital, automated and fast. But, IT solutions must fit into the corporate culture. And they must connect data processing stages. Data from early project phases must be integrated with data from late project phases. Reliable and sustainable data handling is therefore significantly improved.
Thank you, Helen Thomas, for this great contribution.
This shows the case story about AI in pharmaceutical manufacturing from Kasper Larsen, Novo Nordisk. He highlighted that the AI-way is not an easy one.This presentation was part of Track A Session 5 “Artificial Intelligence in Pharma”.
The starting question is:
Kasper Larsen called it “Define & Measure” phase.You shouldn’t only focus on the data scientist role. You also need data engineers, domain experts and business solutions architects/digital translators to make the wheels turn.
Be aware: The balance between the roles will change throughout the phases.
His case is a highly automated packaging line for medicinal pens.
The question is: What is driving efficiency at this line?
They stored the data in a cloud. One important aspect was to bring that data together and have all data available before going to the next step: "Analyse & Improve".
A lot of data have to be collected, and he recommended having a look at real-time and historical data. They used approx. 3 years of data and identified 184 variables.
To predict efficiency, it is important to predict the amount of pens produced per minute.
They found that the biggest impacts are on the following variables:
With this information, they built an AI model, and it could predict the pens/minute. Kasper Larsen proved that the predictions came true.
Thank you for giving us insight into this case out of the practice.
Beginning with deep insights into regulatory projects and challenges. The speakers answered the questions in a very informative panel discussion at the end of the plenary session. This is not a matter of course at many conferences.
Hereafter, three tracks deepened specific areas of regulatory, technology, lab and manufacturing. Good insights, great case studies and inspiring talks. Well done, PDA and all the volunteer speakers, supporters and program planning committee members.
If you haven’t been in Amsterdam this year, be sure to save the date for next year: June 30 - July 1, 2020 in Dublin. And we will be there, too.
See you in Dublin!