Recently a so-called apparent Reynolds number (ARN) approach has been proposed to explain the laminarisation of an isothermal turbulent flow caused by an idealized non-uniform body force. This concept was further applied to heated flows with strong influence of buoyancy, e.g., an upward pipe flow of air or supercritical CO₂. In both cases, turbulent characteristics during the laminarisation process were well explained using the ARN approach. In the present study, this approach (which was only used to explain turbulence) is extended to describing the heat transfer and ultimately for the development of correlations for the Nusselt number in mixed convection flows. The ARN approach is applied to predict heat transfer via two ways: (i) buoyancy is estimated from the DNS data; (ii) assuming the buoyancy is linearly proportional to the Grashof number based on an observation that the non-dimensional temperature profiles are very similar in all shear-driven mixed convection flows studied. Both methods are able to predict the heat transfer deterioration trends, and the former agrees better with the DNS data than the latter. The input of the new correlation is only the apparent Reynolds number (and eddy viscosities from unheated reference flows) and the predictions are obtained numerically without solving the buoyancyinfluenced flow. In future, this new understanding/correlation will be applied to supercritical fluid flows with heating to establish a better understanding on the turbulent heat transfer of supercritical fluid flows.