In the biomedical field, the implants need to be welded with biocompatible materials to adapt different environments in the human body. Laser transmission welding (LTW) is a new method which can be used to connect biomedical materials. However, using the experimental way to get the optimum process parameters of biomedical materials is a time-consuming and high cost project. A numerical simulation-driven experiment design modeling and optimization approach is presented. The LTW of biomedical materials is systematically investigated. Firstly, finite element modeling (FEM) is used to simulate the LTW process and the simulated results is confirmed with welding experiments. Then, an experiment design based on the FEM results is conducted and the artificial neural network (ANN) is used to establish the mathematical models between the process parameters and welding results. The predicted results of the ANN models are tested by FEM. Finally, desirability function (DF) integrated with developed non-dominating sorting genetic algorithm-II (NSGA-II) is used to carry out the multi-objective optimization of the process parameters. It demonstrates that the predicted results of the optimization are in good agreement with the simulated results and experimental results each other, so this approach provides a new way to guide the welding experiments, enhance the welding quality and reduce production cost in the biomedical sector.