Umetrics Suite Blog

Using Data Analytics to Accelerate COVID-19 Vaccine Development

July 23, 2020

In the midst of a global COVID-19 pandemic, a top priority for many pharma and biopharma companies is to get a vaccine developed, produced and delivered to the public as quickly as possible. Ushering a vaccine through rigorous testing protocols and regulatory approvals is not an easy (or quick) effort, but incorporating advanced data analytics could help accelerate the process. Data analytics has proven effective in speeding vaccine development both by enabling more efficient Design of Experiments (DOE) and by creating rapid-scale production rollout processes.


the role of data analytics in COVID-19 vaccine development

The key to accelerating vaccine development in the face of COVID-19 could be to involve data analytics, both using DOE for a more efficient QbD approach and MVDA for faster data insights. 


COVID-19 has presented the biopharma industry with a number of unique challenges. First and foremost, as a novel coronavirus, it’s a new pathogen in humans which has not only made it a challenge to diagnose and treat, but also makes it difficult to develop a vaccine with demonstrated efficacy.  In addition, many of the technologies being pursued for treatment, such as RNA vaccines, are relatively novel themselves. So, while vaccine developers are moving as quickly as possible, the process still must be very thorough. That is where data analytics comes in to play.

Some of the major vaccine challenges that can be addressed with data analytics include:

  • accelerating process development
  • developing robust scale-up and tech transfer methods
  • ensuring manufacturing success improving manufacturing processes

Process Development

When it comes to accelerating process development, Design of Experiments (or DOE) is a tool that allows for a systematic approach to process development studies – ultimately reducing the number of experiments needed, and in the long run, also reducing the overall cost of experimentation. We also can’t overlook the importance that acceleration of process development can have on gaining a competitive edge through speed to market. Afterall, there are more than 150 companies engaged in the race to develop a COVID-19 vaccine candidate.

Support Scale-Up and Tech Transfer

The second challenge that data analytics can support is scale-up and tech transfer. When companies need to quickly and efficiently produce hundreds of millions of doses globally, it's important to have an efficient and organized way to manage scale-up and technology transfer. For any manufacturer, commercial success depends on being able to increase drug substance production volume quickly and effectively and to move to production freely. The time and financial cost of failure can be significant.

The expectation to scale-up to manufacturing in six to 12 months to address COVID-19 is an unprecedented speed. However, by using data analytics tools like MVDA (multivariate data analysis), the number of total batches needed to prove robustness can be less. Therefore, it's important to use the best data you have for single or multiple batch runs during both process development and GMP manufacturing.

Continuously Improve Manufacturing

The last challenge is being able to continuously improve the manufacturing process. Using real-time analytics to monitor and control manufacturing processes has been a proven tool to ensure both process robustness as well as the product quality – even in such expedited timelines.

Addressing major vaccine challenges with Data Analytics

A Closer Look at Accelerating Process Development

During vaccine process development, product components, in-process materials, final product specifications and manufacturing processes are all defined. And at the same time researchers are having to decide on and consider things like safety, efficacy, costs, transport, storage, administration, doses, and immunity.

A Quality by Design (QbD) approach to Design of Experiments (DOE) enables vaccine developers to systematically determine the individual and interactive effects of various factors that influence the results of experiments.

For vaccine development, DOE can be broken down into three different investigational objectives:

  1. Accelerating screening
  2. Supporting optimization
  3. Ensuring robust characterization

Accelerating screening means that vaccine developers can investigate many process parameters at the same time, enabling faster time to market, reduced costs for experimentation, and overall maximized knowledge. One example of accelerated screening can be done during clone screening processes.

The second area where DOE supports accelerated vaccine development is in process optimization. That means developers can use DOE to help determine factors, ranges and inputs needed to achieve specific process goals, for example, to ensure high-quality, high-performing and safe products. One example of this could be media optimization or formula optimization.

The last point is ensuring robust characterization. DOE can be used to analyze each unit operation’s design space, and then to calculate the extent of all the ICH Q8 guidelines. Doing proper bioprocess characterization ensures product stability, robustness and scalability, as well as staying in compliance with regulatory requirements.

How DOE supports vaccine development

So how can we use DOE to improve (and accelerate) vaccine development? The hallmark of Design of Experiments is that this tool or procedure will help you prepare a set of representative experiments in which all factors that you'll want to investigate are varied simultaneously and systematically.

From a set of experiments, you can derive a model that captures the relationship between the factors (settings) and the experimental results (responses).

Design of Experiments (DOE) data analytics


Design of Experiments (DOE) maximizes the information content while keeping the number of experiments low.



In many cases, when running experiments, you’re trying to determine which factors, or what we might call critical process parameters, have an effect on the outcome, and how these parameters might interact with or affect each other.

What you are doing through data analytics is creating a systematic and efficient way of determining which process parameters will have any effect on the critical quality attributes (CQA) or key performance indicators. The way you check this is through data analytics – using statistics to calculate one or more regression models telling you which critical process parameters have any influence on the CQAs or the key performance indicators.

And then, in order to visualize, evaluate and interpret the results, you look at different plots and graphs representing and conveying the essential information. Here (figure 1 below) you can see a number of different plots. The red and green one in the lower right is one of the most essential plots. This is called a design space plot.

examples of plots an graphs for DOE

(Figure 1) Examples of plots and graphs for DOE.

In the figure below (figure 2), you can see the design space plot amplified in more detail. This is a toolset within the MODDE DOE software from Sartorius that helps you visualize the design space and interpret it. The notion of a design space is an essential ingredient in the Quality by Design paradigm that is increasingly advocated or even required by regulatory agencies.

Design space plot (a tool in MODDE)

(Figure 2) The design space plot (a tool in MODDE DOE software) allows you to investigate if there exists a region in factor space in which all goals for the responses are met.


Visualizing and representing a design space this way can help you identify and quantify your critical process parameters and understand exactly which factors will have an effect on the outcome or cause process deviations.

Some Applications for DOE

One example for how DOE is applied toward vaccine development is in evaluating the performance of different cell platforms for viral vaccine process.

The introduction of cell culture systems for virus production has led to major advances in the development of virus-based vaccines. Cell-based culture platforms provide robust cell growth and production of high titers of infections virus particles. It also provides the manufacturer with flexibility in terms of seeding density, density at the time of infection, and harvest of the virus.

In the case of a pandemic, it is critical to increase vaccine production for new virus strains. Cell-based systems can significantly decrease the amount of time it would take to manufacture a new vaccine strain. This is where DOE comes into play.

Use Case: Evaluate Cell Platforms using DOE

MODDE DOE was used to investigate the performance of three different cultivation models for adherently growing Vero cells (T-Flasks, Microcarrier, FibraCel discs) vs. suspension cultivation. The goal of the experiment was to determine which cell platform maximized the viral titer and yield.

Design space of SCD and MOI

(figure 3): Design space for an experiment determining cell platforms


A systematic approach using DOE proposed the best experimental setup as well as described the design space for the following investigations:

  • Optimization for thermal inactivation for LCMV
  • Determination of best sampling time set points
  • Finding the optimal MOI and SCD for adherent cells
  • Evaluation of cell growth inhibiting factors in microcarrier cultivation

evaluation of cell growth inhibiting factors in micro carrier cultivation

The results from the 2-full factorial screening design of inhibiting factors showed a clear negative influence of the residual cell detachment reagent Accutase on the cell propagation. Twenty-four hours post-seeding the model suggested that all of the model terms show a significant influence on the cell density.

The expected beneficial effects of reduced volume and intermittent stirring during seeding could not actually be seen on this DOE. But ultimately, the cells in which Accutase were removed prior to seeding started to propagate earlier and reached an, on average , two to three-fourths higher cell density over the course of cultivation. So, this example shows a really good application in which a DOE experimental setup can be done, and then also how to interpret the results.

Benefits of using DOE

This use case also clearly showed the benefits of using DOE, including:

  • Being able to limit the number of experiments needed to reach conclusions
  • Expediting the process development studies
  • Having users guided toward optimal settings

MVDA provides faster time to insight 

In addition to streamlining your experimental processes for vaccine development using DOE, you can use Multivariate Data Analysis (MVDA) to gain a faster time to insight. MVDA provides a powerful data summary using interactive plots for an intuitive visualization.

Some common applications of MVDA include:

  • Analysis of data originating from spectroscopic measurements
  • Quantitative assessment of process comparability
  • Root cause analysis studies
  • Raw material characterization


Learn more


Want to learn more about how MVDA can be applied to your vaccine development process?

Watch this webinar that covers a use case for evaluating the performance of different cell platforms in DOE. The webinar also provides a case example of MVDA from GKS Vaccines using digital twins to create a vaccine process, plus an example from Intravacc in which SIMCA helps optimize and accelerate process development for antigen concentration for a SIPV process.

Digital twins are the next evolution of machine learning, which combines data analytics and simulation with comprehensive models that blend historical information with real-time data to predict the future of a process. (Read more about using digital twins in biopharma).


Watch the Webinar



Topics: Multivariate Data Analysis, Design of Experiments (DOE), Pharmaceutical manufacturing, Medicine/Health

Tiffany McLeod

Written by Tiffany McLeod

Tiffany McLeod is the (Bio)pharma Market Manager within the Sartorius Stedim Data Analytics marketing team. She has been employed by Sartorius since 2017. Within this function Tiffany acts as the teams’ subject matter expert for life science market trends and requirements. She also works to addresses and develop data analytics solutions to solve industry challenges. She is passionate about biopharma 4.0 and helping businesses pursue digital transformations. Tiffany holds a degree in Bioengineering and Bioinformatics from the University of California, San Diego.

Leave a comment

Search the Blog

    Subscribe to the Blog

    View the:

    Data Analytics Glossary of Terms

    List of Webinars

    Get a free trial