Machine Learning Enhances Therapeutic Antibodies: Study
Washington, 12 Sep (ONA) --- Scientists at the University of Michigan have unlocked a new frontier in drug development with the application of machine learning to fine-tune therapeutic antibodies, which have immense potential in treating conditions like Parkinson's, Alzheimer's, and cancer.
Antibodies can be powerful disease fighters, but they often bind to unintended molecules, reducing their efficacy. The University of Michigan's machine learning models can pinpoint these problematic areas in antibodies. What sets them apart is their adaptability - they can be used on existing antibodies, those in development, and even ones that haven't been created yet.
The researchers found that many clinical-stage antibodies fail this balance, with 75% interacting with the wrong molecules or with each other. The machine-learning models, trained on clinical-stage antibody data, significantly expedite this process by identifying modifications with up to 88% accuracy.