The authors aimed to study the skin surface bioheat perfusion model described in part I numerically. The influence of each constituent in the determination of surface temperature profile was statistically examined. The theoretically derived data will then be benchmarked with clinically measured data to develop the artificial intelligence system for the diagnosis of erectile dysfunction (ED). The new approach is based on the hypothesis that there exists a constitutive relationship between surface temperature profiles and the etiology of ED. By considering the penis model as a group of reservoirs with irregular cavities, we built a numerical model, simplified to save computational costs while still realistically able to represent the actual for partial differential calculation. Incompressible blood flow was assumed coupled with the classical bioheat transfer equation which was solved using the finite element method. Isotropic homogeneous heat diffusivity was assigned to each tissue layer. The results of simulations were tested for sensitivity analysis and further optimized to obtain the 'best' signal from the simulations using the Taguchi method. Four important parameters were identified and analysis of variance was performed using the 2 n design (n ¼ number of parameters, in this case, 4). The implications of these parameters were hypothesized based on physiological observations. Our results show that for an optimum signal-to-noise (S/N) ratio, the noise factors (thermal conductivity of skin, A and tunica albuginea, B) must be set high and low, respectively. Hence, at this setting, the signal will be captured based on the perfusion rate of the boundary layer of the sinusoidal space and the blood pressure (perfusion of sinusoidal space, C and blood pressure, D) will be optimal as their S/N ratios (C (low) and D (low)) are larger than the former.