Blog

Improving Quality Control – the Cornerstone of Pharma Manufacturing 

pharma-header-vaccine-1210x242

It doesn’t take a mental giant to understand the crucial importance of quality control in the Pharmaceutical Manufacturing industry. Besides having to meet strict local and international regulations and quality standards for products made for human use and consumption, there are many reasons why quality control is the cornerstone of this industry.

For example, the quality of the raw materials used in the manufacturing have to meet specific standards along with the status and health of manufacturing equipment/assets and processes. And let’s not forget the manufacturing environment which must meet extremely high clean/sterile specs. There are also the training and qualifications of the personnel who oversee the plant and production.

As you can see, if anything goes wrong in any of these areas, the material/product will most likely have to be trashed which can amount to high loss and high costs. Therefore, the pharmaceutical manufacturing industry must ensure it has a secure handle on quality control which means it makes good business sense to invest in digitalization and using all technology at its deposal to accomplish this goal.

However, according to the 2019 McKinsey Global Institute Report, the pharmaceutical manufacturing industry is lagging way behind in this area and in fact is at the very bottom of the list of industries that have embarked on their digitalization journeys and that have also embraced the accompanying technologies that make the most out of process manufacturing data.

One such technology is self-service analytics, and there is an excellent argument to be made for using this tool in the pharma manufacturing industry, especially for quality control.

Making the Most Out of Production Data 

pharma-lab-tech-quality-control-500x333Pharmaceutical processes capture a huge amount of time-series data but also have tons of contextual data such as material information, lab tests and results, quality checks, and so on. To optimize operations, it makes sense to make the most out of this data.

In fact, it makes sense to integrate these two types of data and by doing, so you as plant managers and process experts can gain a deeper understanding and a deeper visibility into your processes and production. And this translates into important business value for your organization.

So, why don’t you use self-service analytics so you can improve operations but most importantly so you can improve quality control.

Here’s what you’ll be able to do:

  • Make captured time-series data ready for analytics through plug & play indexing of all historical data.
  • Use any tag for analytics, e.g., pattern searches, diagnoses, comparisons, and monitoring.
  • Use fingerprints, conditions, and business rules for setting early warnings and soft alarms to capture extra events and send out notifications via text messages or emails to alert personnel about process issues.
  • Introduce asset awareness by using an asset structure directly.
  • Contextualize time-series data by adding contextual information as discrete or sequential events keeping process experts and records up to date.
  • Visualize these context items on diagrams with search, comment, and collaborative events.
  • Stay aligned with existing workflows and processes by integrating alerts with 3rd party tools thus eliminating data silos.
  • Establish production dashboards so all personnel have “eyes on production”.

Use cases are great because they give you a solid example and explanation of how a tool work. Let’s look at one, so you can directly understand how self-service analytics can help you ensure quality control and optimize production.

Use Case: Quality Control for Vaccine Production   

During production of a COVID-19 vaccine, process experts observed a quality problem. Using self-service analytics, they investigated the issue by analyzing the process data. They were able to visualize the details of production trends and other derived process information to understand which factors were affecting the batches. Below, you can see the investigative process these experts followed.  

Phase 1: Issue Assessment  

The team first used their custom production dashboard to look at trends which showed that the cycle time of a recent batch was abnormally long. Additionally, the box-plot tile within their dashboard showed that some of the process phases had higher variations compared to others. They next looked at the monitor and “current value” tiles to see how the process had been behaving at current points in time. It showed that these were within the expected control parameters. Then, zooming into a tile on the dashboard containing contextual process information, they were only concerned with specific sequential “failed batch” events to find the root cause of the low-quality material.  

TrendMiner screenshot

Figure 1. Failed Batch Event Details. 

Phase 2: Root Cause Analysis   

pH is a major factor in final product quality, so the process experts started with this parameter. Using the Recommendation Engine of the self-service analytics, a feature that uses machine learning to generate recommended solutions/answers to process questions, they saw a drop in temperature that seemed to be causing a drop in pH. They thought this was most likely the cause of low-quality product but had to check it out. They compared a failed batch against a set of previously approved batches (which had been saved as a “golden fingerprint”) to see where the process had deviated. This comparison showed that when the temperature dropped out of threshold, the pH had also dropped out of threshold, proving that their hypothesis was correct.  

TrendMiner screenshot

Figure 2. Bad Batch vs. Vaccine Golden Batch. 

The shaded hull regions are from a fingerprint that was produced using the golden batches. The solid red, blue, and orange lines represent the bad batches which clearly drift outside the hull regions established by the golden fingerprint. 

Phase 3: Proactive Measures   

It’s not enough to only find a root cause, the issue needs to be resolved so that future issues are prevented. And this is what the process experts did. The team created a monitor that could be used with the golden batch fingerprint to detect future deviations from their ideal batch thresholds. In case a potential process deviation is detected, personnel would be alerted giving them enough time to take action.  

TrendMiner screenshot

Figure 3. Monitor Vaccine Quality.   

Greater Rewards for Success through Self-Service Analytics  

In referring to the current industrial manufacturing situation, the McKinsey Global Report mentioned above states that “In this world, the rewards for success—and the penalties for failure—are ever greater”. It goes on to elaborate on insights for manufacturing industries concerning their digitalization journeys and the adoption of new technology which will bring substantial benefits and help them maneuver towards the future.   

The pharmaceutical industry will gain substantial business value using self-service analytics. It can greatly help the industry and you, its plant managers and process experts, with the job at hand and in particular help with its process cornerstone: improving quality control.