In this paper, a novel method is proposed for the determination of the optimal subject-specific placement of knee implants based on predictive dynamic simulations of human movement following Total Knee Arthroplasty (TKA). Two knee implant models are introduced. The first model is a comprehensive 12-degree-of-freedom (DoF) representation that incorporates volumetric contact between femoral and tibial implants, as well as patellofemoral contact. The second model employs a Single-Degreeof-Freedom Equivalent Kinematic (SEK) approach for the knee joint. A co-simulation framework is proposed to leverage both knee models in our simulations. The knee model is calibrated and validated using patientspecific data, including knee kinematics and ground reaction forces. Additionally, quantitative indices are introduced to evaluate the optimality of implant positioning based on three criteria: balancing medial and lateral load distributions, ligament balancing, and varus/valgus alignment. The knee implant placement is optimized by minimizing the deviation of the indices from their userdefined desired values during predicted sit-to-stand motion. The method presented in this paper has the potential to enhance the results of knee arthroplasty and serve as a valuable instrument for surgeons when planning and performing this procedure.