In this study, we developed and validated an electromyography- (EMG) assisted musculoskeletal simulation framework with concurrent optimization of knee kinematics and muscle excitations. The musculoskeletal model had a 12 degree of freedom (DoF) knee joint with personalized articulating surfaces. First, model?s muscle parameters underwent calibration, followed by the EMG-assisted analysis. To assess model?s performance, we compared estimated knee biomechanics against four other simulation approaches, i.e., a 12 DoF knee model with either 1) uncalibrated EMG-assisted and 2) static-optimization (SO) neural solution; and a conventional 1 DoF knee model with either 3) EMG-assisted or 4) SO neural solution. The performance of the models was assessed against in vivo measured values from two grand challenge datasets. For estimated muscle excitations and joint contact force (JCF), the EMG-assisted models outperformed the SO solutions. Compared to the EMG-assisted 1 DoF knee, using EMG-assisted 12 DoF knee improved estimation of muscle excitations, joint moments, and transverse tibiofemoral JCF to a greater extent than compressive tibiofemoral JCF. To estimate compressive tibiofemoral JCF (during walking), the EMG-assisted model with personalized 1 DoF knee may suffice. However, the EMG-assisted 12 DoF knee model is recommended for a more accurate estimation of joint moments, muscle forces, and compressive and transverse tibiofemoral JCF, especially when these quantities can be affected, e.g., due to musculoskeletal disorders. The developed simulation framework provides a viable approach for estimating knee biomechanics accounting for personalized muscle excitation strategy and knee articulating geometries.