Written by Moe Elbadawi.

Over the past several years, we have demonstrated how 3D printing enables on demand access to personalised medicines; promising a new era for patient care. In response to emerging technologies, we have recently adopted machine learning into our 3D printing workflow, facilitating us to accelerate the research and development of novel formulations. This union of machine learning with pharmaceutical 3D printing is a world first. Machine learning, a subset of AI, is a state of the art technology that facilitates accurate prediction of experimental outcomes without the need for physical experimentation in the laboratory. For this reason, it can save time, resources, and optimise the development of new 3D printed medicines.

In our M3DISEEN study, we apply machine learning to predict optimal parameters for the production of medicines with fused deposition modelling (FDM). Using our algorithm we predict key FDM processing parameters with an accuracy of just under 80%, essentially reducing the time needed to fabricate novel formulations. In addition, we are able to predict both the hot melt extrusion and FDM temperatures required to produce distinct formulations with an error of less than +/-10 °C.

Typically, building machine learning algorithms requires a lot of data. The data used to build our M3DISEEN algorithm was carefully collected by UCL and FabRx 3D printing experts over a 6-year period. We acknowledge that this may not be feasible for every research group in which researchers are pressed for time in collecting the data. Fortunately, there are emerging machine learning algorithms suited to low datasets, which we discuss in our recent review titled 'Advanced Machine Learning Techniques in Drug Discovery'.

For us, M3DISEEN is only the beginning of our work in applying artificial intelligence to the 3D printing of medicines. In the near future we hope to expand on our portfolio to illustrate how the cutting-edge partnership can be harnessed to increase the translation of 3D printing into clinical settings.