The complexity of biologic drug manufacturing and distribution introduces the potential for variation that could affect patient safety, yet little published literature exists studying the consistency of clinical outcomes at a large scale for biologics. Using public, lot-level adverse event (AE) reports for two major biologics and four major vaccines from the FDA’s Adverse Event Reporting System, we estimate competing Hidden Markov Models (HMMs) to infer the most probable structure of the stochastic process generating observed AEs per lot. We find evidence of infrequent but persistent latent states with high AE risk, indicating dependence between AE counts for subsequent lots. Though we cannot yet rule out alternative explanations, these results are consistent with the hypothesis that manufacturing and distributions conditions effect AE report frequency. 

Many powerful tools for process data analytics and artificial intelligence (AI) have been developed in the last decade. This  presentation describes the application of such AI tools to
biopharmaceutical manufacturing. In particular, an approach to automated data analytics is described in which an expert system (aka “decision tree”) automatically selects the best methods based on a systematic interrogation of the characteristics of the dataset. This smart process data analytics approach is demonstrated in several biomanufacturing case studies.

This presentation will cover the development of a fully automated and integrated continuous downstream platform with an overarching automated control system. This system of modular construction can be rapidly reconfigured and easily adapted for processing most products. Following on from successful demonstration of the system’s capability, expanded digital integration in the addition of real time model predictive/prescriptive maintenance is being implemented to further improve efficiencies in biologics manufacturing.

Biologics process development architecture with real-time analytics is one of the key components of a successful roadmap for biomanufacturing 4.0 and real-time release initiatives. Realtime monitoring of biopharmaceutical development process entails strategic deployment of analytical technologies, and cyber physical systems for visualization & systematic data management. While in-line and on-line analytical sensors facilitate instant acquisition of data from bioprocesses; integration of automated data analysis, visualization and repository infrastructures enable holistic process monitoring and understanding. 

The current focus on data-centric process development and optimization drives the awareness to purely data-driven digital approaches. However, these approaches easily fall short in unraveling the underlying system behavior. We propose to use management systems of existing biological knowledge to contextualize data, allowing to understand the mechanisms that govern specific phenotypes acquisition across organisms and conditions. We show that the development of systematic solutions for knowledge management is the key for impactful digital transformation of bioprocessing.

The presentation will address central challenges in digitalization and big data analytics in biopharma and will demonstrate the potential to systematically provide value through integration of smart digital technologies into the work stream. The presentation will be based on several industrial use cases in USP and DSP showing benefits from software-assisted and -enabled process monitoring, control, optimization and automation.

Digital twins are supposed to be the key enabler for knowledge management and the digital transformation of biomanufacturing. This contribution focuses on strategies to automatically derive hybrid digital twins and even more important to adapt them along the product life cycle to provide a platform technology. We will show case studies demonstrating multiple benefits of that approach: Optimization of productivity, streamlining process characterization and tech transfer acceleration.

Biochemical changes essential for cell growth occur within the sub-cellular spaces, for which there are no readily available technologies. Fluorescent nanosensors have the potential to address this challenge, due to their biocompatibility, accuracy and real-time measurement capabilities. They have already demonstrated their potential to quantify pH, molecular oxygen and temperature in complex model systems, such as the nematodes, fungi, stem cells and electronic devices. Therefore, exciting findings could be translated to augment cell and gene therapy manufacture.