AI-Based Data-Efficient Approach for Formulation Screening and Dissolution Modeling

In vitro dissolution testing plays a key role in controlling the quality and optimizing the formulation of pharmaceutical products. Traditional approaches to screening and developing dissolution models require extensive experimental trials, which are resource-intensive and time-consuming. This session presents an AI-based, data-efficient method leveraging active learning and Gaussian processes to streamline this process. By integrating these advanced ML techniques, we aim to demonstrate the utility of these approaches for the selection of formulations with desired dissolution characteristics while reducing the need for exhaustive testing. Active learning strategically selects the most informative experiments, thereby enhancing data efficiency, while Gaussian processes provide a probabilistic framework that captures uncertainty in predictions. This approach not only accelerates the formulation development cycle but also improves the accuracy of dissolution models, ultimately facilitating more effective translation from laboratory to clinical applications. Initial results demonstrate the potential of this methodology to significantly cut down on experimental costs and time, paving the way for more sustainable and innovative pharmaceutical practices.