The study of infectious diseases and their spreading mechanisms is fundamental to prevent new infections and limit the eventual death toll. Just as not all diseases have the same characteristics, not all individuals have the same probability of contracting the infection and different scenarios lead to different outcomes. Focusing on the heterogeneity of the populations, we create an interdisciplinary and novel approach that relax the assumption of well-mixed populations, typical of compartmental models, by incorporating state-of-the-art results of empirical and analytical pedestrian studies. By knowing the relationship between crowd density and walking velocity we are able to devise a density-dependent contact rate allowing us to estimate the number of contacts between infectious and healthy individuals in crowded and confined situations. The advantages of this method are: (i) the ability of getting more realistic solutions on the number of new infections and (ii) the possibility of studying a scenario a-priori. This means that just by knowing the size of the population and the extension of the environment we are able to infer the number of contacts between individuals and, consequently, the number of new infections. We apply the method to the London Underground network by using data from Transport for London (TfL). We firstly devise a transportation model that evaluates the amount of time passengers spend in each station in order to walk from entrance to platform and vice-versa. After that we infer the density inside the stations and the number of contacts between passengers. As a validation we compare the results with data of infections (reported by general practitioners) in London boroughs and obtain a clear correlation between the use of public transport and the incidence rate of infections. Further studies using this new type of modelling could help evaluate and, eventually, prevent contagion in most crowded environments such as public transport, offices, schools and hospitals.