Learning to simulate realistic human diffuse reflectance spectra

Neural surrogate models can achieve MC-level accuracy and <i>in vivo</i> realism at negligible inference cost, enabling large-scale, compute-efficient data generation for biomedical optics and robust AI development for clinical applications.