This thesis introduces a new approach to characterize and evaluate ocular light exposure based on the discovery of novel blue light-sensitive photoreceptors in the human eye. These photoreceptors are the primary mediators of 'non-visual' responses impacting human health, from resetting the circadian clock to directly alerting the brain. In recent years, studies have found that light at short-wavelengths is more effective than light at longer wavelengths at inducing and suppressing a range of 'non-visual' responses. Although it has been recommended that we approximate the spectral sensitivity of these novel photoreceptors with an action spectra peaking near 490 nm, the optimal approach for quantifying non-visual spectral effectiveness is yet unknown. These novel photoreceptors, in addition being photoreceptors themselves, receive inputs from the classical photoreceptors (rods and cones) that in return affect the overall spectral sensitivity of the non-visual system as it changes with lighting conditions. Due to this time-varying spectral sensitivity and the relatively slow temporal processing, the relation between dynamic external light stimuli and the magnitude of non-visual responses cannot be explained with a single function or a simple threshold value. To better understand these non-linear and unknown relations, this thesis aims to develop a novel computational method, based on recent findings about the 'non-visual' - also called non-image-forming - effects of light on human health. A dynamic wavelength-dependent model framework is proposed to evaluate the non-visual health potential of light. This novel approach integrates the spectral effectiveness of irradiation and accommodates time-varying spectral sensitivity functions. These time series of light quantities serve as inputs for the light-driven model, which accounts for light intensity, duration, history, and timing of light exposure. By quantifying light in terms of spectral effectiveness and temporal dynamics, different light exposure patterns can be ranked in terms of its potential to have an impact on human health. The final objective of this thesis is to support the design of healthier buildings by applying evidence-based lighting criteria, which can then inform architectural design through a simulation-based approach. Incorporating non-visual effects into a building simulation workflow requires a good approximation of daylight spectra as it varies with sun position and sky type. Performance predictions must also account for occupant behavior and scheduling, which brings us to the question: How can we apply such a method to make informed decisions about our built environment? The integration of the proposed model into a functional simulation workflow is demonstrated using an architectural case study but first the non-visual spectral effectiveness of light will be evaluated under varying sky conditions to analyze the model output sensitivity to input accuracy. The model and its application to the built environment will then be investigated using multiple view directions and by considering occupant behavior and scheduling to make an immersive prediction within a space. This novel computational approach can be seen as a first step towards human-centric lighting application, simulating an occupant's light consumption to evaluate non-visual health potential that can support decision-making in the built environment.