Molecular imprinting, also referred to as template polymerization, is a method of preparation of materials containing recognition sites of predetermined selectivity . Biomimetic assays with molecularly imprinted polymers (MIPs) could be considered as alternatives to traditional immuno-analytical methods based on antibodies. This is due to a unique combination of advantages displayed by MIPs including synthetic procedure that does not require animal inoculation and sacrifice, conjugation of hapten to a carrier protein for stimulated production, the possibility of manufacturing MIPs against toxic substances, excellent physicochemical stability, reusability, ease of storage, and the ability to perform recognition in organic media .
The conventional method for preparing MIPs is bulk polymerization  followed by grinding and sieving to obtain appropriately sized particles for further use. These are irregular and polydisperse and usually include a large portion of fine particulate material. Extensive sieving and sedimentation are required to achieve a narrow size distribution and to remove fine particles which make this method time consuming and labor intensive. Moreover, the obtained polymers have many limitations, including a high level of nonspecific binding and poor site accessibility for template molecules and therefore are not used in commercial assays.
New methods of MIP synthesis in the form of micro- and nanoparticles offer better control of the quality of binding sites and morphology of the polymer. Micro- and nanostructured imprinted materials possess regular shapes and sizes and a small dimension with extremely high surface-to-volume ratio with binding sites at close proximity to the surface . This greatly improves the mass transfer and binding kinetics. These factors are very important for facilitating binding and improving sensitivity and speed of sensor and assay responses.
Recently, we have developed the first prototype of an automatic machine for solid-phase synthesis of MIP nanoparticles using a reusable molecular template . The instrument for the production of MIP nanoparticles consists of a computer-controlled photoreactor packed with glass beads bearing the immobilized template. It can be suitable (in principle) for industrial manufacturing of MIP nanoparticles. The feeding of monomer mixture, reaction time, and washing and elution of the MIP nanoparticles are under computer control which requires minimal manual intervention. The broad range of parameters which can vary during synthesis of nanoparticles requires extensive optimization of manufacturing protocol. In our work, the composition of monomer mixture is selected using the computational approach developed earlier, which has proven its efficiency and become routinely used in many laboratories worldwide . However, the synthesis of MIPs is a process involving several variables. Its optimization is still a complex task due to the interconnected nature of factors that influence the quality and yield of MIPs .
For this reason, the optimization of synthetic conditions by one-variable-at-a-time (OVAT) is unsuitable and cannot guarantee that real optimum will be achieved. The OVAT approach is only valid if the variables to be optimized are totally independent from each other . With the rapidly rising cost of making experiments, it is essential that optimization is done in as few experiments as possible. This is one important reason why statistical experimental design is needed. Design of experiments (DOE) originated as a method to maximize the knowledge gained from experimental data. Compared with conventional methods, multivariate approaches based on DOE allow studying all possible interactions between experimental variables and can significantly reduce the experimental effort needed to investigate the experimental factors and their interactions. These methods are especially valuable for optimization of chemical processes. The examples of application of multivariate DOE include using MODDE 6 software for optimization of supercritical fluid extraction, conditions for the extraction of indole alkaloids from the dried leaves of Catharanthus roseus, and GC/MS-based analysis of amino acids and organic acids in rat brain tissue samples [9, 10]. Only a few reports discussing the chemometrics approach in rational design of MIPs have appeared. Thus, Kempe and Kempe  employed multivariate data analysis (MODDE 6.0 software, Umetrics, Umea, Sweden) for the optimization of monomer and cross-linker ratios in the design of a polymer specific for propranolol. Mijangos et al.  used chemometrics (MODDE 6.0 software, Umetrics, Sweden) to optimize several parameters such as concentration of initiator (1,1′-azobis(cyclohexane-1-carbonitrile) and 2,2-dimethoxy-2-phenylacetophenone) and polymerization time required for the design of high-performance MIP for ephedrine.
In the present work, we demonstrate the use of the multivariate DOE approach and MODDE 9.0 software (Umetrics, Sweden) for increasing the yield of MIP nanoparticles synthesized in the automatic photoreactor developed by our team.