Open-pored lattice structures can be designed and adapted in a variety of ways to be used in many fields of application and contribute to the optimisation of existing components. However, the large number of different shapes and customization options also makes it difficult to select the optimum lattice structure. For this reason, the "KIKI" project is working on the development of an AI-supported design workflow for the generation of bio-inspired lattice structures that can be used to structure mechanical test specimens and, in the long term, to improve additively manufactured endoprostheses made of Ti-6Al-4V. AI-supported lattice development is used to select the optimum lattice configuration for a given load case and enables a stress-optimized design. In this way, the customizability of endoprostheses, the most natural possible load distribution between bone and prosthesis, and the ingrowth of bone can be improved. To train the AI, the first step is to create a data base. For this purpose, a wide variety of diatom-based lattice structures are developed. Once suitable structures of the diatoms have been identified, parametric unit cells are derived from them, which are developed and optimized in algorithm-based design software. Subsequently, regular, uniform, open-pore lattices are generated from the unit cells. The choice of unit cell, a customizable uniform material thickness, and a possible rotation of the unit cell can be used to adapt the generated lattice. These design principles are developed in such a way that they can be used in a highly variable manner, taking design properties relevant to endoprosthesis into account and enable optimization in terms of weight, porosity, strength, and producibility. Further, a suitable AI model is developed to obtain optimal parameters for lattice structure generation. Through sophisticated algorithms and neural networks, it analyzes the data base of over 40,000 examples to determine the most suitable lattice configurations for user inputs. This tool supports the selection of suitable lattice structures in the design process of the bio-inspired mechanical samples and thus improves the generation of an optimized structure. The AI tool's ability to adapt and learn from big data base ensures precise outcomes tailored to specific requirements, marking a significant advancement in the use of optimized lattice structures in engineering and health care sector. Keywords: Deep Learning, Optimization, Health Infrastructure