Artificial Intelligence-Driven Quality by Design in Pharmaceutical Formulation Development: Applications, Challenges, and Future Perspectives

The development of pharmaceutical formulations is changing. It is moving away from trial-anderror ways. Now modern approaches rely on science and data. Quality by Design or QbD offers a structured way to develop medicines. It starts with defining the Quality Target Product Profile or QTPP. Then it identifies Critical Quality Attributes or CQAs. It also looks at Critical Material Attributes or CMAs and Critical Process Parameters or CPPs. Design of Experiments or DoE helps in planning studies. This leads to the creation of a design space and a control strategy. Traditional QbD methods may not work well when variables are complex. They may not handle effects or strong interactions between factors. Artificial intelligence machine learning and predictive models can help. These tools can find patterns in data from experiments, formulations and manufacturing. This review looks at how AI fits into every stage of formulation development. It covers preformulation excipient selection, compatibility prediction, formulation optimization and predicting CQAs. It also includes identifying CMAs and CPPs. AI helps with process monitoring, Process Analytical Technology or PAT design space development and control strategy design. Applications are shown in solid dosage forms, modified-release systems, nanoparticulate and lipid-based formulations, continuous manufacturing, personalized medicines and 3D-printed medicines. Some challenges exist. These include data, overfitting models, models that are hard to understand lack of validation difficulties in scaling up regulatory issues and the need for diverse expertise. New ideas, like twins, continuous process verification, hybrid models that combine mechanics and AI generative AI and federated learning may push the field forward. In the end AI should support science not replace it. Experimental verification and mechanistic understanding are still essential.