**5. Digitalization**

Biologics manufacturing, like many other manufacturing industries, is being influenced by the Industry 4.0 push. However, from the numerous concepts currently explored under the digitalization umbrella, which spans across multiple sectors, three key advances are likely to impact the biologics industry and create further incentives to adopt continuous manufacturing. These are smart sensors, Big Data, and Digital Twins.

#### *5.1. From Smart Sensors to Big Data*

Smart sensors, as discussed in Section 3.2, as part of the general digitalization drive, are indeed a promising tool to obtain large and improved datasets that better reflect the state of development of a process. Leveraging these data is the other half of the digitalization movement. In this case, the objective is to take the large amount of data collected and convert it into actionable information. Initially, the aim is to monitor a process with the intention of later using this information to change/control the process [65]. In batch production, these concepts have already been employed to predict batch end-time from seemingly information-poor variables, such as temperature and pH. Concepts such as predictive maintenance also leverage data to predict and schedule maintenance operations [65].

#### *5.2. Digital Twins*

Like Big Data, Digital Twins form one of the key pillars in Industry 4.0 and the digitalization drive [23,49].While the Digital Twins concept is somewhat vaguely defined for bio-based manufacturing, there are two types of digital twins that can aid future development. This first type of digital twins deals with operational support and control. For example, the use of digital twins to forecast the evolution of a fermentation process, such as in [66]. On the other hand, a digital twin can be a digital representation of a future production process, where it acts as a validated test-bed which can be used to refine and build confidence in a process design prior to construction.

All these developments directly impact further strengthening of the case of continuous production and its business case. The development of novel sensors enables collecting datasets that are needed for "real-time" tracking of key state variables in continuous production. Simultaneously, the big data-based process monitoring and control methods are required to ensure that the data gathered can be turned into actionable information for the process operators and/or perform closed-loop control action. Digital twins, on the one hand, play a similar role in producing actionable information and providing the ability to implement closed-loop controls.

A knock-on effect of the operation supports that these elements provide, together with the high degree of process automation, is needed for continuous production processes. This brings the need for reduced staff for given production output. In addition, the operators who are working on continuous production processes would operate the process through more automated operations, avoiding manual tasks as much as possible, which are the norm in batch production.

#### *5.3. Application of Modeling in Regulatory Decision*

A unique and challenging aspect of introducing changes to pharmaceutical processes' design and operation is the need for regulatory bodies to approve any specific changes. To this end, a concept such as "Digital Twins", which, in essence, opens up towards shifting validations and testing on a process into an in-silico environment, needs to be accepted by the regulatory bodies. Quality by design is one such framework. It has been adopted and endorsed by the FDA, which indicates their willingness to acknowledge the need for more in-silico-based studies to improve the precision and speed at which process designs can be created and tested. However, the question remains whether multiple designs (process paths) can be validated by employing the digital twin concept. At the very least, these digital twins can be applied for building a multivariate design space and scale-down models for commercial-scale systems, reducing the economic burden on the R&D department without compromising the quality.

#### *5.4. Leveraging Process Data*

With the development and pilot-scale operation of hundreds of process designs in a year and dozens of commercial production processes, pharmaceutical companies can collect a large amount of diverse process data from operations. With the recent advances made in data-driven analytics and the availability of "big data", pharmaceutical companies can leverage these data to identify common failures and successes rapidly. In turn, these situations can be further analyzed by subject matter experts to develop process insights for an improved design and operation of future production processes. As such, data analytics will allow pharmaceutical companies to rapidly improve process designs as opposed to the more subdued pace at which changes usually occur. It is noteworthy that data

analytics acts as an enabler for experts to analyze hundreds, if not thousands, of datasets effectively, thus facilitating the gathering of insights and pro-active planning of operational changes.

#### **6. Hybrid Facilities: Acknowledging the Best of Both Worlds**

Depending on the demand and manufacturing stage, it becomes easier or more challenging to work with single-use/disposable systems or multi-use stainless steel systems. A hybrid facility that uses disposables and reusable systems potentially combines both systems' benefits and could be the path forward depending on the required capacity. This has often come to mean single-use technologies in seed-train development. However, for upstream production, the use of stainless steel-based perfusion bioreactors is more common, and then again using single-use systems in downstream processing for in-process holding and filtration units. Such flexibility will automatically enable the flexibility of connected unit operations and, thus, continuous operations while keeping the risk of contamination to a minimal level [36].

#### **7. Summary and Outlook**

Despite the benefits of continuous over batch bioprocessing, its adoption has lagged, with few exceptions. However, the batch manufacturing paradigm's dominance in the industry for reasons such as "batch processing is familiar and works very well" cannot be sustained in the long term given the new biomanufacturing challenges. The industry-held perception of complexity in continuous bioprocessing is becoming obsolete, as more and more new technologies and solutions are continually improving the situation. Several academic- and industry-led consortia are working to improve the perception regarding continuous bioprocessing by bringing the questions to the correct stakeholders who can address them. The training provided by these initiatives to the top managemen<sup>t</sup> of the companies is playing an essential role in changing the perception and, at the same time, also creating new scientists and operators that can understand and respond to a new set of operational challenges. However, wider adoption of continuous bioprocessing will only be possible if the gaps at the technical, management, and regulatory levels are acknowledged. As discussed in this paper, concerted efforts are being made to abridge them. These include:
