Abstract:This paper addresses the challenges of predicting bioprocess performance, particularly in monoclonal antibody (mAb) production, where conventional statistical methods often fall short due to time-series data's complexity and high dimensionality. We propose a novel Hyperbox Mixture Regression (HMR) model which employs hyperbox-based input space partitioning to enhance predictive accuracy while managing uncertainty inherent in bioprocess data. The HMR model is designed to dynamically generate hyperboxes for input samples in a single-pass process, thereby improving learning speed and reducing computational complexity. Our experimental study utilizes a dataset that contains 106 bioreactors. This study evaluates the model's performance in predicting critical quality attributes in monoclonal antibody manufacturing over a 15-day cultivation period. The results demonstrate that the HMR model outperforms comparable approximators in accuracy and learning speed and maintains interpretability and robustness under uncertain conditions. These findings underscore the potential of HMR as a powerful tool for enhancing predictive analytics in bioprocessing applications.

What problem does this paper attempt to address?

### Problems the paper attempts to solve This paper aims to address the challenges in performance prediction during monoclonal antibody (mAb) production. Specifically: 1. **Complex time - series data**: Traditional statistical methods often perform poorly when dealing with high - dimensional time - series data because these data have complex internal correlations, leading to inaccurate prediction results. 2. **High - dimensional data space**: The monoclonal antibody production process involves a large number of input parameters, and the high - dimensional nature of these parameters makes it difficult for traditional methods to handle effectively. 3. **Model transparency and interpretability**: In industrial applications, especially in the biopharmaceutical field, models not only need high precision but also need to be able to explain their prediction results for practical operation and decision - making. 4. **Uncertainty management**: Biological process data usually contains measurement noise and other uncertain factors, so the model needs to be able to reason and interpret under uncertain conditions. To meet these challenges, the authors propose a new model based on Hyperbox Mixture Regression (HMR). The HMR model divides the input space by dynamically generating hyper - boxes, thereby improving prediction accuracy, learning speed, and maintaining model transparency and robustness. This model is particularly suitable for predicting Critical Quality Attributes (CQAs) and Key Performance Indicators (KPIs) in the monoclonal antibody production process. ### Main contributions 1. **Introduction of a new neuro - fuzzy model structure**: The HMR model combines hyper - box fuzzy sets and local linear regressors and can work effectively in high - dimensional data spaces while maintaining model transparency. 2. **Efficient one - pass learning process**: The HMR model uses a one - pass input data transfer process for feature space partitioning, significantly improving the learning speed. 3. **Dynamically weighted combination of local linear regressors**: By dynamically weighting and combining the local linear regressors associated with each hyper - box, the accuracy of the model is improved and the network complexity is reduced. 4. **Normalization layer**: A normalization layer is introduced to reduce the risk of numerical instability in subsequent layers. 5. **Performance prediction for the next two days**: The HMR model can predict key performance indicators in the monoclonal antibody production process for the next two days, such as Viable Cell Density (VCD) and mAb concentration. ### Experimental results The experimental results show that the HMR model outperforms the existing ANFIS Hybrid Learning (HL) and Fuzzy Neural Network (FNN) Back - Propagation (BP) algorithms in both high - dimensional and low - dimensional scenarios. Specifically: - **High - dimensional scenario**: When using all 23 input features, the HMR model significantly outperforms the ANFIS HL and FNN BP algorithms, especially with a lower Root Mean Square Error (RMSE) on the test set. - **Feature selection**: Through the feature selection algorithm, the HMR model can further improve the prediction performance while reducing the number of input features and the model complexity. In conclusion, the HMR model proposed in this paper provides an efficient, transparent, and robust solution for performance prediction in the monoclonal antibody production process.

Hyperbox Mixture Regression for Process Performance Prediction in Antibody Production

Model-based Process Development of Continuous Chromatography for Antibody Capture: A Case Study with Twin-Column System

Converting a Mab Downstream Process from Batch to Continuous Using Process Modeling and Process Analytical Technology.

A Robust Statistical Batch Process Monitoring Framework And Its Application

Uncertainty Quantification Using Ensemble Learning and Monte Carlo Sampling for Performance Prediction and Monitoring in Cell Culture Processes

Prediction of antibody production performance change in Chinese hamster ovary cells using morphological profiling

A Data-Driven Approach for Leveraging Inline and Offline Data to Determine the Causes of Monoclonal Antibody Productivity Reduction in the Commercial-Scale Cell Culture Process

A deep learning-aided multi-objective optimization of a downstream process for production of monoclonal antibody products

Application of advanced machine learning algorithms for anomaly detection and quantitative prediction in protein A chromatography

Chemometric Identification of Canonical Metabolites Linking Critical Process Parameters to Monoclonal Antibody Production During Bioprocess Development

Data-driven prediction models for forecasting multistep ahead profiles of mammalian cell culture toward bioprocess digital twins

Improving N-Glycosylation and Biopharmaceutical Production Predictions Using AutoML-Built Residual Hybrid Models

Application of Raman spectroscopy in monoclonal antibody producing continuous systems for downstream process intensification

Adaptive local weighted kernel-based regression for online modeling of batch processes

Digital Twin in High Throughput Chromatographic Process Development for Monoclonal Antibodies

Recursive Gaussian Process Regression Model for Adaptive Quality Monitoring in Batch Processes

A Novel Incremental Gaussian Mixture Regression and Its Application for Time-varying Multimodal Process Quality Prediction

Multiphase and Multimode Monitoring of Batch Processes Based on Density Peak Clustering and Just-in-time Learning

Data-driven and Physics Informed Modelling of Chinese Hamster Ovary Cell Bioreactors

Probabilistic model by Bayesian network for the prediction of antibody glycosylation in perfusion and fed‐batch cell cultures

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