Related Experiment Video
Updated: Aug 12, 2026
07:37
An Integrated Raman Spectroscopy and Mass Spectrometry Platform to Study Single-Cell Drug Uptake, Metabolism, and Effects
Published on: January 9, 2020
Deep Learning Algorithm-Guided Raman Spectroscopy for Real-Time Correlation of Critical Process Parameters and Titers
Meiqi Shi1,2, Yuan Liu3, Haiyuan Chen1,2
1Department of Pharmaceutical Engineering, School of Biomedicine and Health Sciences, Beijing Institute of Petrochemical Technology, Beijing, Beijing, China.
Biotechnology and Bioengineering
|August 10, 2026
Summary
This study introduces an advanced deep learning model for real-time monitoring of lentiviral vector production. The ACNN model accurately predicts critical parameters, improving gene therapy manufacturing efficiency.
Area of Science:
- Biotechnology
- Process Analytical Technology (PAT)
- Gene Therapy Manufacturing
Background:
- Lentiviral vector (LV) production for gene therapy faces challenges in scalability and cost-efficiency due to slow offline monitoring of critical parameters and viral titer.
- Real-time monitoring is crucial for optimizing LV production processes and ensuring product quality.
Purpose of the Study:
- To develop a process analytical technology (PAT) for real-time monitoring of LV production using Raman spectroscopy and a novel deep learning algorithm.
- To enable accurate prediction of glucose, lactate, viable cell density, and viral titer during LV production.
Main Methods:
- Integration of data augmentation, self-supervised learning, and time-series modeling for Raman spectral analysis.
- Development of an advanced Convolutional Neural Network (ACNN) model incorporating Squeeze-and-Excitation (SE) attention, relative positional encoding (RPE), multi-query attention (MQA), and Top-K feature selection.
- Utilized unlabeled spectral data to extract features from complex components, including viral particles.
Main Results:
- The ACNN model significantly outperformed the traditional Partial Least Squares (PLS) model in predicting viral titer, cell density, glucose, and lactate concentrations (p < 0.05).
- Substantial reduction in viral titer prediction error was achieved with ACNN.
- ACNN demonstrated significantly lower RMSE and higher R² for titer prediction compared to PLS (p < 0.01 and p < 0.001, respectively).
Conclusions:
- This study overcomes technical bottlenecks in real-time quality monitoring and process optimization for LV production.
- The developed ACNN model provides a feasible transition towards transparent and intelligent monitoring in LV manufacturing.
- Establishes a foundation for large-scale intelligent production of gene therapy vectors.
