Opportunities for Digital Twins in Bioprocess Development
The global COVID-19 pandemic has accelerated the adoption of digital transformation across pharma and biopharma companies around the world – with a simultaneous move toward technologies based on machine learning and Artificial Intelligence (AI), including the creation of digital twins to provide in silico replicas of structures, processes, and instruments.
This article is posted on our Science Snippets Blog.
Digital twins rely on advanced data analytics to establish replicas of platform processes that can be used to optimize process development and process validation.
Digital twins represent the next generation of process development, embodying the push toward Pharma 4.0 while embracing the advanced data analytics tools essential to digitalizing the full bioprocess lifecycle development. Digital twins rely on advanced data analytics to establish replicas of platform processes that can be used to optimize process development and process validation.
What Is a Digital Twin?
A digital twin is a comprehensive, integrated mathematical model based on historical and current data that simulates a complete bioprocess sequence in real-time. It incorporates Process Analytical Technology (PAT) data, quality data, and time-series data, and can help predict critical quality attributes (CQA) and key performance indicators (KPIs). A digital twin creates a simultaneous replica of an existing real-world process in order to predict the outcome in advance.
A digital twin is the next evolution of digital modeling using virtual simulation combined with process control strategies. For pharma companies, perhaps the most exciting benefit of using a digital twin is being able to simulate process deviations and understand the resulting effects, without having to spend months or years in scale production or testing.
The global market for digital twins has been estimated to be worth $3.1 billion and is expected to reach more than $48 billion by 2026.
Digital Twins Support Bioprocess Validation
In the pharmaceutical industry, bioprocess validation is an essential step in the commercial development of any drug, as well as for FDA and other regulatory agency approvals. Process validation is one of the key application areas for digital twins, and in fact, was one of the first uses. By reducing the number of process performance qualification (PPQ) runs and enabling a robust control strategy, digital twins can save manufacturers millions in development costs.
Digital twins support process validation in a number of ways:
- As part of process characterization studies, digital twins help establish the acceptance criteria and normal operating ranges, which aligns with the FDA’s Quality by Design approach.
- Incorporating digital twins during PPQ helps biopharma companies reduce the number of PPQ runs needed for validation.
- Digital twins help biopharma companies establish alert limits, predict deviations and react to trends during continued process verification (CVP).
Ensuring GMP-Compliant Manufacturing
In the biopharma industry, Good Manufacturing Practice (GMP) – ensuring that products are consistently produced and controlled according to quality standards – is essential not only to ensure products are safe but also to meet regulatory compliance standards. Digital twins support biopharma GMP in a number of ways.
- Establish more accurate control models. Setting the specifications for manufacturing in the most robust manner ensures fewer out-of-spec (OOS) events. Simulations based on digital twins are better able to predict variations and help establish more precise control specifications and control limits, which leads to fewer deviations and OOS events.
- Predictive control models. Digital twins based on advanced data analytics tools can seamlessly predict the future based on past data. Combining historical data, present data, and advanced models within a simulation-based on multivariate data analytics (MVDA) offers the most precise models of complex bioprocesses.
- Continuous process improvement. By incorporating machine learning algorithms into the model and integrating new production data, digital twins can improve in accuracy over time.
- Faster problem-solving. With the ability to drill down to uncover the root cause of events and simulate variations, digital twins based on MVDA tools help manufacturers identify and correct process deviations quickly.
Creating a Digital Twin
The primary consideration for creating a bioprocess digital twin is being able to accurately leverage data from process development, design of experiments (DOE), multivariate data analysis (MVDA), and domain knowledge about the process. The setup requires a seamless digital collection of data from processes into a digital storage database, statistical analysis to create control strategies, and automatic modeling for ongoing processes.
Nearly 50% of companies still primarily use Excel or even paper to record data—even if they’re partially digitized.
Data from cellular environments, changing conditions of processes, instruments, and workflows all need to be measured, recorded, and analyzed. This creates a large amount of data that must be evaluated and understood. Being able to ascertain the effect that interconnected conditions and variables have on each other requires the use of multivariate data analytics tools (MVDA), such as SIMCA® and SIMCA®-online for real-time monitoring.
A recent survey by Deloitte with MIT Sloan found that only 20 percent of biopharma companies are digitally maturing. That could be why nearly 50% of companies still primarily use Excel or even paper to record data. And even among companies that are partly digitized, data is often kept in data silos or not fully integrated among departments.
This shows a clear need for biopharma companies to shift away from old systems of data storage and management and move toward digital platforms designed to integrate and analyze digital data in user-friendly ways. Adopting digital collect and storage systems, and data analytics tools such as MODDE® for DOE, SIMCA® for MVDA, and SIMCA®-online for real-time process monitoring help companies make the shift toward digital transformation and realize the benefits of using digital twins.
Examples of Digital Twins in Bioprocess Development Space
Bioprocess digital twins have the potential to simulate sensor measurements – also known as “soft-sensors”, process units, or even connected process units. Within the bioprocess development space, the advancement of digital twins has 3 main drivers:
- Minimizing the number of physical experiments needed to be run, therefore reducing development effort and COGS
- Reducing the time needed to run multiple experiments in vivo, therefore shortening time to market
- Maximizing the number of process scenarios that can be considered when developing a process, therefore increasing process and product understanding
Digital twins can be applied to both upstream and downstream process development scenarios.
Within the upstream process development space, digital twins can, and have been, used to simulate biochemical pathways for cells, simulating processes like biomass accumulation (i.e.VCD, dead, and lysed cell concentrations), metabolic activity (i.e. glucose metabolism), metabolic flux, as well as glycan profile formations. These simulations can be applied to designing optimal control loops for substrate additions (i.e. nutrients and inducers) as well as to predict the cellular activity for different feed-strategy methods (i.e. perfusion vs. fed-batch).
Upstream digital twins are typically building with the “hybrid” approach to modeling, which utilizes mechanistic and material balance “rules” combined with empirical process data.
Within the downstream process development space, digital twins are fairly more established as they tend to rely more on mechanistic approaches and follow natural laws from the fields of fluid mechanics, kinetics, and thermodynamics. Downstream digital twins can be used to simulate chromatography as well as filtration processes. These simulations can be applied to estimate the optimal loading conditions, expected impurity concentrations, as well as processing times.
Learn More
Find out more about how Amgen is using data analytics and digital transformation technologies to improve processes and predict problems before they occur.