Abstract:In biopharmaceutical manufacturing, fermentation processes play a critical role in productivity and profit. A fermentation process uses living cells with complex biological mechanisms, leading to high variability in the process outputs, namely, the protein and impurity levels. By building on the biological mechanisms of protein and impurity growth, we introduce a stochastic model to characterize the accumulation of the protein and impurity levels in the fermentation process. However, a common challenge in the industry is the availability of only a very limited amount of data, especially in the development and early stages of production. This adds an additional layer of uncertainty, referred to as model risk, due to the difficulty of estimating the model parameters with limited data. In this paper, we study the harvesting decision for a fermentation process (i.e., when to stop the fermentation and collect the production reward) under model risk. We adopt a Bayesian approach to update the unknown parameters of the growth-rate distributions, and use the resulting posterior distributions to characterize the impact of model risk on fermentation output variability. The harvesting problem is formulated as a Markov decision process model with knowledge states that summarize the posterior distributions and hence incorporate the model risk in decision-making. Our case studies at MSD Animal Health demonstrate that the proposed model and solution approach improve the harvesting decisions in real life by achieving substantially higher average output from a fermentation batch along with lower batch-to-batch variability.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to optimize the fermentation harvest decision - making with limited data in the biopharmaceutical manufacturing process. Specifically, the paper focuses on: 1. **Purity - Yield Trade - off**: During the fermentation process, the target protein and unwanted impurities grow simultaneously, which leads to a trade - off between purity and yield. In order to obtain a high protein content, decision - makers may be inclined to harvest the fermentation products as late as possible. However, waiting too long will lead to an increase in the impurity level, thereby increasing the difficulty and cost of subsequent purification operations. Therefore, how to determine the optimal harvest time to maximize the expected profit is a key issue. 2. **Process Uncertainty**: Uncertainties in the fermentation process are divided into two categories: intrinsic randomness and model risk. Intrinsic randomness means that due to the complexity of the biological system, even when fermentation is carried out under the same conditions, the accumulation rates of proteins and impurities are random. Model risk refers to the uncertainty introduced by the difficulty in accurately estimating model parameters due to limited data. These uncertainties bring additional challenges to the harvest decision - making. 3. **Decision - Making with Limited Data**: In the research and development stage and the early production stage, usually only limited historical data are available. In this case, how to develop a learning mechanism to make harvest decisions while considering intrinsic randomness and model risk is an important research problem. In summary, the main research problem of the paper is: in the case of limited data and process uncertainty, how to formulate the optimal fermentation harvest strategy to maximize the expected profit obtained from a batch of fermentation products and reduce the variation between batches.

Biomanufacturing Harvest Optimization with Small Data

Production Optimization for Concentration and Volume-Limited Fed-Batch Reactors in Biochemical Processes

A Robust Statistical Batch Process Monitoring Framework And Its Application

Bioprocess Modeling Using A Robust Genetic Programming

Interpretable Biomanufacturing Process Risk and Sensitivity Analyses for Quality-by-Design and Stability Control

Stochastic Simulation Model Development For Biopharmaceutical Production Process Risk Analysis And Stability Control

[New Opportunities and Challenges for Hybrid Data and Model Driven Bioprocess Optimization and Scale-Up].

Multi-fidelity Gaussian Process for Biomanufacturing Process Modeling with Small Data

Policy Optimization in Dynamic Bayesian Network Hybrid Models of Biomanufacturing Processes

Stochastic Simulation Uncertainty Analysis to Accelerate Flexible Biomanufacturing Process Development

A New Data-Driven Modeling Method for Fermentation Processes

Stochastic Process Optimization of an Integrated Biorefinery

Pharmaceutical Process Optimisation: Decision Support under High Uncertainty

From Spatial-Temporal Multiscale Modeling to Application: Bridging the Valley of Death in Industrial Biotechnology

Data-Driven Dynamic Modeling and Online Monitoring for Multiphase and Multimode Batch Processes with Uneven Batch Durations

A Two-Phase Approach Optimizing Productivity for a mAb-producing CHO Cell Culture Process Using Dynamic Response Surface Methodology Models

Optimization and Scale-Up of Fermentation Processes Driven by Models

Optimization Models for Integrated Biorefinery Operations

Towards advanced bioprocess optimization: A multiscale modelling approach

A multi-stage stochastic programming model for adaptive biomass processing operation under uncertainty

Optimizing Bioprocessing Efficiency with OptFed: Dynamic Nonlinear Modeling Improves Product-to-Biomass Yield