Raman Band Selection for Accurate Biopharmaceutical Aggregate Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting the concentration of aggregates in biopharmaceutical suspensions, such as those containing antibodies, suffer from low accuracy due to the selection of wave number bands in Raman spectrum measurement data that are not truly contributing to the prediction, leading to unreliable results.
Innovation Solution
An information processing device and method that selects specific wave number bands by comparing intensity values of Raman spectrum data from antibodies and their aggregates, using a machine learning model like a neural network, to generate a state prediction model with higher accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If sparse modeling is used to select wave number bands, then the selection process can be automated, but the selected wave number bands become highly dependent on the specific measurement data and lack generalizability
Solution Approach 1:
The patent performs preliminary action by pre-defining multiple wave number bands associated with specific amino acid residues and protein secondary structures before the actual prediction. This allows the system to automatically select from pre-identified bands without relying on data-dependent sparse modeling, thereby achieving both automation and generalizability. The wave number bands are selected based on predetermined biochemical knowledge rather than fitting to specific measurement data.
2Reliability
If wave number bands are selected based on predetermined biochemical knowledge, then the selection is more scientifically grounded, but the process becomes more complex requiring multiple comparisons
Solution Approach 1:
The patent applies segmentation by dividing the Raman spectrum into multiple discrete wave number bands, each associated with specific amino acid residues or secondary structures. This segmentation allows systematic comparison of intensity values across predefined bands rather than analyzing the entire spectrum at once, reducing computational complexity while maintaining scientific rigor. The segmentation enables structured processing of spectral data according to biochemical knowledge.
3Measurement precision
If all wave number bands are used in prediction, then no selection bias is introduced, but the prediction accuracy decreases due to inclusion of irrelevant bands
Solution Approach 1:
The patent extracts only the relevant wave number bands associated with amino acid residues and secondary structures that contribute to aggregate formation. By taking out and selecting only these specific bands rather than using all available bands, the system achieves higher prediction accuracy while reducing the number of parameters. The extraction process removes irrelevant spectral information that would dilute the predictive signal.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The proposed method allows for more accurate prediction of aggregate concentrations in biopharmaceutical suspensions, enhancing the reliability of manufacturing processes by identifying relevant wave number bands and improving the precision of concentration predictions.
Implementation Method 1
measuring a Raman spectrum of a suspension
Data Source
AI summary
An information processing device includes a processor, in which the processor is configured to: as preparatory processing for generating a state prediction model that predicts a state of a target component in a suspension produced in a manufacturing process of a biopharmaceutical containing a target protein as an active ingredient, acquire first spectrum measurement data obtained by measuring a spectrum of an electromagnetic wave emitted from the target protein and second spectrum measurement data obtained by measuring a spectrum of an electromagnetic wave emitted from the target component; and select a specific wave number band or a specific wavelength band that is specific to the target component by comparing an intensity value of the first spectrum measurement data and an intensity value of the second spectrum measurement data.