Per says predicting your development and production processes is essential for achieving consistency and quality, as well as for avoiding compliance issues (you could lose your license). It’s importance to know what results to expect and why you can expect them.
Per also points out that things will change over time, and by better understanding your processes, you can then adapt to differences you can’t avoid. Per sees how processes can interconnect. He has experienced situations where the production quality issues were a result of product and process development issues. Addressing process improvements upstream leads to overall process improvement.
We also discussed some interesting data challenges. Even for companies that are collecting production data, if they’re not using it, data quality can be very low (no eyeballs on the data at all can leave a lot unnoticed and unfixed). Per has witnessed many pharmaceutical companies wanting to see stability in their production data, so sometimes there is no variation in the data. But this lack of variation impedes a better understanding of production processes. Doing designed experiments makes the variation more evident, enabling companies to truly learn from the data. As Per so eloquently states, “If you want to see the influence of a variable, it has to vary.” This is why Per and his colleagues at NNE are such strong proponents of designed experiments. “Design of experiments is by far the best tool to obtain process understanding,” he says.
With the new Functional Data Explorer in JMP 14, Per talks about some of the many opportunities they are pursuing in biotech with their process data over time, as well as optimizing their legacy products in production.
These are a few of the highlights from our conversation. For more of Per’s wisdom, check out the latest episode of Analytically Speaking.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.