Moving From Data Paralysis to Predictive Analysis

Posted Nov 25, 2016

Read Kierans blog following the ISA/ISPE event on the 22nd of November on Moving From Data Paralysis to Predictive Analysis.

20 years ago Kaizan, LEAN, 5S and 6 sigma were the techniques being applied to improve efficiency, reduce waste and remove variability in processes. We analysed and recorded data and measured our successes, however that analysis tended to be historical and was most effective on processes that were well understood. These techniques still apply to manufacturing facilities today, however now we have added a mountain of data to be recorded, analysed, interpreted and presented in real-time, on processes that have become more complex and are less understood.

The nature of the Pharmaceutical and Healthcare manufacturing Industry is changing. We are seeing an increased number of multi-product process trains, coupled with more complex manufacturing processes in the biotech/biologics market. The challenge for the pharma sector today is to take the mountain of data and apply the correct data analytics tools/techniques to provide a meaningful and useful output. When you have hundreds of potential variables in a process train knowing which ones are the most critical to achieve optimum efficiency and output is difficult.

It is no surprise to see pharma companies and manufacturing facilities at very different stages of the journey, not only in the application and approach but also in terms of the technology, connectivity and availability of data. The recent ISA/ISPE Seminar: “Moving from Data Paralysis to Predictive Analysis” in Cork gave an insight into how companies are approaching this challenge in different ways. It also highlighted that despite different approaches the challenges facing the industry were common.

At an organisation level the main reoccurring challenge is making data analytics a key cross functional goal with senior management buy-in across the organisation. Many speakers identified the need to identify smaller projects and clear organisational wins to help establish a business case and clear ROI for larger projects. At a technical level one of the advantages of using smaller projects before launching an organisation wide project, is that you get to proof out the processes, techniques and systems at a smaller scale.

Another major challenge is the need to change the culture of the company. That is the need to transform users to “Citizen Data Scientist”, moving them from an understanding of univariate analysis to an awareness of multi-interdependent-variable analysis. Data analytics projects need to be driven by these user groups. The “Citizen Data Scientist” not only has an understanding of the process or business functions, but also can identify where data analytics can help solve problems or increase knowledge to optimise processes or workflows. This can be achieved by the use of education, coaching, and mentoring but also by running smaller projects to show people the possibilities.

As the proportion of drugs manufactured using a biotech process increases, the level of risk in our drug supply chain increases due to the complexity, variability and current understanding of these processes. The data captured around these processes hold the key to gaining control and minimizing risk and delivering stable regulatory compliance. Without understanding the operational risk of the hundreds of interdependent variables in the biotech process, a manufacturing site will struggle to maintain efficiency and financial success and runs the risk of falling foul of regulatory authorities.

The world of Smart data has only just begun. Grasp the opportunities!