ML-Based Excipient Selection for Protein Viscosity Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for selecting viscosity-reducing excipients for protein compositions are time-consuming and require prior knowledge of the target protein, limiting their effectiveness in predicting viscosity changes and optimizing formulations, especially when dealing with high concentrations that lead to increased viscosity and aggregation issues.
Innovation Solution
A computer-based method using a dataset of known formulations and Machine Learning models to predict the viscosity-changing effect of excipients on unknown proteins, allowing for the selection of optimal excipient combinations without requiring detailed protein information, thereby reducing the need for costly and time-consuming laboratory tests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Volume of moving object
If high concentration protein formulations are used to reduce injection volume, then the therapeutic dose can be delivered in small volumes, but the viscosity increases significantly making injection difficult
Solution Approach 1:
The patent applies parameter changes by systematically varying excipient types, concentrations, and combinations to optimize formulation viscosity. Machine learning models predict viscosity outcomes for different parameter settings, enabling selection of excipient formulations that maintain low viscosity at high protein concentrations, thus enabling small volume injections while preserving injectability.
2Reliability
If screening approaches are used to identify viscosity-reducing excipients, then the best excipient combinations can be found, but the process is time-consuming and requires extensive laboratory testing
Solution Approach 1:
The patent implements preliminary action by using machine learning models to predict excipient viscosity effects before conducting laboratory experiments. The models are trained on existing data and can screen numerous excipient combinations in silico, identifying promising candidates for experimental validation. This preliminary computational screening significantly reduces the number of wet-lab experiments needed, accelerating the formulation development process while maintaining reliable excipient selection.
3Measurement precision
If prior knowledge of the target protein is required for viscosity prediction, then accurate predictions can be made, but this limits the method's applicability to unknown or new protein therapeutics
Solution Approach 1:
The patent achieves universality by developing machine learning models that can predict viscosity for both known and unknown proteins. The models are trained on diverse protein-excipient datasets and use transfer learning approaches, allowing them to generalize to new protein therapeutics without requiring extensive protein-specific prior knowledge. This enables the same platform to serve multiple protein formulation needs across different therapeutic areas.
4Productivity
If high concentration formulations are used, then the number of injections required can be reduced, but protein aggregation and particle formation increase
Solution Approach 1:
The patent uses excipients as intermediary substances that mediate between high protein concentration and protein stability. Specifically, surfactants and stabilizing excipients are selected to prevent protein aggregation and particle formation at high concentrations. The machine learning models predict not only viscosity but also stability outcomes, enabling selection of excipient combinations that maintain protein stability while allowing high concentration formulations for reduced dosing frequency.
Data Source
AI summary
Method for selecting at least one viscosity changing excipient for a formulation containing at least one unknown protein via a computer includes providing a data set that describes the viscosity of several known formulations containing at least one protein and optionally at least one viscosity changing excipient; generating representations of at least one excipient; using a Machine Learning Model executed on the computer to recognize patterns in the data set to evaluate the viscosity changing effect of the viscosity changing excipient to a new formulation containing at least one unknown protein by applying the recognized patterns on provided data of the unknown protein; selecting the at least one excipient according to an acquisition criterion and applying the excipient to the unknown protein, wherein the provided data of the at least one unknown protein describe the viscosity of a protein composition containing the at least one unknown protein.