The development of robust protocols and the establishment of a products are the essential building blocks upon which scientists advance knowledge and technology in any field of inquiry. This includes the field of nanomedicine. In particular, the creation of nanoparticle-based drug delivery systems (DDSs) involves many different experimental conditions that could influence the outcomes and that largely depend on the specific material in study.

Any unknown process can be thought of as a “black box” in which the operator introduces materials and choses specific settings. The process returns some measurable outputs (Fig. 1A). Thus, process optimization requires both understanding which settings (or parameters or variables) are relevant, and among these, select the best level of each to achieve a desired output. To achieve this, it is necessary to correlate the variables with the outcome features (Fig. 1B). When trying to optimize the features of these nanovectors, researchers are presented with the dauting task of understanding the significance and effect of many experimental factors in the process output. This hurdle induces scientists to use two main approaches.

The first one is a trial-and-error strategy, in which each variable is tuned singularly, selecting the best outcomes before optimizing the next one (Fig. 2B). This approach requires a limited number of experiments and can lead to good results. However, this strategy presents intrinsic limitations. Firstly, it is not possible to understand if some variables have more leverage than others, and no information is acquired on the possible synergistic or antagonistic effects of multiple variables interactions. Secondly, the final nanovectors may not be the best possible ones in the considered experimental range, but just a “local optimization” that is derived from the narrow scope of the screening.

The opposite approach to trial and error consists a complete screening of all the conditions in the selected experimental range (Fig. 1C). This strategy allows to acquire complete knowledge on the process. However, it can require a very high number of experiments to optimize even a handful of experimental factors, and becomes exponentially more expensive to perform when increasing the number of considered parameters and their different levels. This can require a very high amount of resources and workforce that can outweigh the benefits deriving from an optimal outcome.

Design of Experiments (DoE) tries to bridge the gap between these two opposite approaches (Table 1). DoE is a statistical methodology based on the simultaneous tuning of experimental parameters. It allows to create an optimal set of experiments that provides the maximal amount of information on the process, depending on the objective of the study. Thus, DoE allows to compromise some level on information to significantly reduce the time and resources allocated to the understanding and optimization of any given process (Fig. 2D). This powerful tool can be used efficiently in research, when the aim is to quickly select only variables with a significant effect on the final outcomes and to optimize them to achieve the desired results. This is especially relevant in research groups with limited workforce and resources.

Despite its high potential, DoE is often overlooked outside of the engineering field, and its effective application and correct interpretation can result challenging to non-specialists due to the complex mathematical theory it involves. This is demonstrated the very small percentage of articles published in PubMed that apply DoE in their studies. Although, this percentage increased in the last twenty years, it still represents only 2% of the overall publications on nanomedicine (Fig. 3).

In this review, we offer a practical guide on the use of DoE to optimize nanovectors for drug delivery. We will focus on elucidating DoE terminology, discuss recent and most relevant examples of DoE application to nanomedicine, and give some practical advice on how to apply DoE to any nanoparticle study, including future directions. Despite not being an extensive discussion on DoE, we believe this article can give the basic tools to understand DoE to a wider public, and communicate the potential it holds for discovery and innovation, and give a primer on how to implement it, together with possible sources for more in depth DoE exploration.