Pressure ulcers are a significant concern in nursing care. However, preventing pressure ulcers often relies solely on visual assessments of skin tissue. For example, changes in skin color can be due to factors such as oxygen deficiency and external effects. However, the internal composition of skin may vary significantly despite appearing identical on the surface. Given that skin tissue is semitransparent to light, long-wavelength colors such as red and yellow penetrate deep into the skin tissue and reflect at different depths. This phenomenon provides spectral images representing the superimposed state of the tissue layers. These overlapped reflection spectra are undetectable by the naked eye and can only be identified by hyperspectral imaging. In this study, we introduce the double-end generative matching technique to address the insensitivity in detecting skin abnormalities by sight alone. Unlike traditional methods in which RGB images are used, we believe that the correct estimation of hemoglobins must retrieve the complete composition of chromophores through the reflected hyperspectral lights within layers of skin. This method uses a Monte Carlo method for multilayered media (MCML) simulations to produce non-overlapping surface spectra for specified damaged tissue, with authentic images providing feedback to refine the simulated skin model’s accuracy. We generate a vast training dataset for bidirectional training with the trained optical model, and we can accurately predict the concentration of skin chromophores to identify subtle differences that may not be visible to the naked eye. This method enables the identification of areas in the dermis with low levels of oxygenated hemoglobin (Hb) and abnormal water content. From an arbitrary picture, our model decomposes a complete set of contents into skin layers. Two chemical substances are used to predict ulcers, and the rest of the paper will be used to validate the consistency of our estimations. Our results show significant consistency between our prediction and the ground truth.