Three-dimensional (3D) printing emerged as an advanced technology for rapid prototyping, enabling the production of intricate objects layer-by-layer under the precision guidance of computer-aided design (CAD) models. The unique manufacturing process of 3D printing offers many advantages and is particularly suitable for the manufacture of personalized, complex structures, and has made significant progress in many fields (Hu et al., 2021, Urciuolo et al., 2020). In pharmaceutics, 3D printing exhibits revolutionary potential, particularly for solid oral dosage forms (Ragelle et al., 2021, Seoane-Viano et al., 2021). It allows for variations in size and geometry to control the dose and release behavior with characteristics that cannot be achieved in conventional dosage forms, and is one of the key technologies to realize personalized treatment and customized medicine (Wang et al., 2022). Khaled et al. used 3D printing to manufacture solid dosages containing five compartmentalized drugs with two independently controlled release profiles, which can improve patient adherence and allow for the ready tailoring of a particular drug combination or drug release to meet individual needs (Khaled et al., 2015). Ghanizadeh et al. designed and used 3D printing to create a bilayer tablet containing two tuberculosis drugs, isoniazid, and rifampicin, which could be released at a suitable pH to provide better clinical outcomes (Ghanizadeh Tabriz et al., 2021).

Despite the significant advances of 3D-printed pharmaceutics for polypill and controlled release, the design of composition and geometry for dosage forms has not kept pace with technological advances. The rapid development of 3D printing technology offers an unprecedented ability to control the composition and structure of every voxel (Skylar-Scott et al., 2019), which means it is possible to manufacture personalized drugs that meet individual needs and precisely control drug release. However, traditional experience-based solid dosage geometry design is not suitable for such complex structures, and cannot give full play to the advantages and characteristics of 3D printing. It is necessary to propose a new design paradigm for the accurate geometry structure design of solid dosage forms. Some pioneer research works have been done for 3D printed solid dosage structure design. The basic drug release behavior of 3D-printed solid dosage forms was studied, and the results show that drug release depended on surface area to volume (SA/V), and was controlled by erosion-mediated. Further studies have shown that the mean dissolution time (MDT) of printed drugs can be predicted using (SA/V), and it is effective for both water-soluble and poorly water-soluble pharmaceutical substances as well as for erodible and inert polymers. (Goyanes et al., 2015, Thanawuth et al., 2023, Windolf et al., 2021). The understanding of the drug release behavior promotes the prediction of theoretical drug release profiles based on complex structures, and some programmed release profiles as delayed, pulsed, or constant can be designed and achieved with core–shell, multilayer, and gradient structures (Haring et al., 2018, Tan et al., 2020). Unfortunately, in contrast to the ease of obtaining the release profile for certain dosage forms, the precise design of the dosage form with the desired release profile is very difficult and can be seen as a complex optimization problem in mathematics (Grof and Štěpánek, 2021).

Artificial intelligence (AI) is a technology that simulates human intelligence processes, enabling computers to achieve autonomous learning, reasoning, perception, and decision-making. AI has shown amazing potential in several fields in recent years (Castro et al., 2021, Gainza et al., 2023). For medical 3D printing technology, its automatic and digital characteristics make it a natural partner of artificial intelligence (Elbadawi et al., 2021a, Elbadawi et al., 2021b). The application of artificial intelligence technology will effectively assist the formulation development process of drug 3D printing, build automated design and manufacturing solutions for personalized medicine, and meet the drug requirements of different patients efficiently and at a low cost, as shown in Fig. 1. Elbadawi and Ong et al. used machine learning techniques to develop and update the M3DISEEN software, which can predict the printability of different preparations in the pharmaceutical field (Elbadawi et al., 2020, Ong et al., 2022). Obeid et al. used a self-organized graph (SOM) and multilayer perceptron (MLP) model to study the effects of SA/V ratio and printing parameters on diazepam release behavior(Obeid et al., 2021). Recently, Elbadawi et al. proposed a method utilizing conditional generative adversarial networks (cGANs) to create novel formulations for 3D printing in medicine development, achieving a balance between creativity and realism, and demonstrating the potential of AI to revolutionize the drug development process.(Elbadawi et al., 2024). These studies combining 3D printing with artificial intelligence provide valuable references for materials selection and printing parameter optimization. Regarding dosage geometry forms design, Mazur et al. attempted to use artificial neural networks (ANN) to predict the optimal geometry corresponding to the required dose and release curve; however, their test set accuracy was only 44.4 %, indicating that this area needs further exploration (Mazur et al., 2023).

Compared to other algorithms, the genetic algorithm in AI is a well-known metaheuristic optimization algorithm inspired by the process of natural selection and is particularly well-suited for geometry structure optimization (Kanyilmaz et al., 2022, Katoch et al., 2020). However, despite its potential in personalized medicine through the optimization of drug delivery systems, the integration of genetic algorithms with advanced manufacturing techniques, specifically 3D printing, remains a relatively unexplored territory. In this work, we explored combining genetic algorithms and 3D printing techniques to design and manufacture custom capsules with optimized structures and desired drug release profiles. Firstly, the dissolution simulation model was set up and verified by the dissolution test. After that, the genetic algorithm was used to design and optimize the capsule geometry structure according to the specific drug dissolution curve. Finally, it is found that 3D printing can accurately manufacture capsules designed by genetic algorithm, and the dissolution behavior of printed customized capsules has a good consistency with the target curve (f2 value is over 50), which is expected to be used to guide more accurate 3D printing personalized drug design.