Making in vitro release and formulation data AI-ready: A foundation for streamlined nanomedicine development

Journal article, 2025

Machine learning and artificial intelligence (AI) is transforming the way pharmaceutical products are developed across drug discovery, process engineering, and pharmaceutics functions. AI for nanomedicine development is enabling faster and more accurate prediction of critical quality attributes (CQAs). However, the full potential of AI is limited by the quality and accessibility of data. Unlike adjacent fields such as the chemical sciences, the pharmaceutics domain lacks curated, open-access databases, particularly for nanomedicines. To address this, here we curate an open-access local database focused on liposomal formulations. The database includes formulation parameters, in vitro release (IVR) testing conditions, and digitised drug release data. By evaluating the entries in the database qualitatively and quantitatively, we identified challenges in current data reporting practices. This includes incomplete reporting of formulation and IVR testing conditions, as well as inconsistent quality of drug release plots and their data format. Based on our analysis, we propose a set of data standards and a database structure to support harmonisation for nanomedicine formulation and IVR data. Our open-access database aims to improve data accessibility and transparency to enable the development of robust AI models for IVR and CQA prediction, ultimately streamlining nanomedicine development.