Electron-Cyclotron waves are routinely used in tokamaks to heat the plasma and to drive current through the electron channel. Their unique ability to drive very localized current renders them the main foreseen tool for neoclassical tearing mode mitigation in future large fusion devices, such as ITER. This application is critical to avoid dangerous disruptions, which can damage the tokamak components, e.g. the vacuum vessel or the coils. However, experimental characterizations of the wave power deposition have led to the conclusion that numerical simulations, using either ray-tracing or quasilinear drift-kinetic codes, tend to predict an overly narrow power deposition profile and to overestimate the current drive efficiency, as well as the total driven current. This thesis focuses on understanding the discrepancy between experiments and simulations, to improve the predictive capabilities of the numerical toolkit. In particular, efforts have been placed in investigating the interplay between the injected waves, the suprathermal electrons and the missing ingredient in the codes usually used to simulate electron-cyclotron wave propagation and absorption: turbulence. Two phenomena are proposed to explain this discrepancy: the beam broadening caused by its scattering through density fluctuations, and the wave-enhanced transport of suprathermal electrons, transporting electrons away from the resonance location. A first part of the thesis is dedicated to the study of turbulent transport enhancement by the electron-cyclotron wave absorption. The wave-accelerated electrons emit hard X-rays through Bremsstrahlung, which can be measured in TCV using the Hard X-Ray Spectrometer. It is then possible to perform forward modeling of a discharge thanks to the synthetic diagnostic embedded in the quasilinear bounce-average drift-kinetic Fokker-Planck solver LUKE. It is shown, by analyzing experimental data, that the radial transport of fast electrons increases with the injected wave power. This is further confirmed by matching experimental and synthetic hard X-ray data, tuning an ad-hoc radial diffusion coefficient in LUKE. This increase of transport is associated with a measured increase of density and temperature fluctuation level. A new electron-cyclotron heating and current-drive source has been developed for the global flux-driven gyro-kinetic code ORB5, simulating turbulent transport from first principles and showing an increase of electron transport with wave power. On the other hand, beam scattering is also studied, using a full-wave propagation solver, coupled to LUKE to estimate the impact of beam broadening on the wave power deposition profile. Dedicated scenarios have been developed in TCV to minimize and maximize this impact, respectively. Even though the impact of beam broadening is expected to be important in large fusion devices like ITER, it is shown that, for the tested TCV configurations, this phenomenon is not sufficient to explain the gap between experimental and synthetic hard X-ray data. This thesis provides tools and an analysis framework for a better understanding of the underlying mechanisms behind the suprathermal electron distribution broadening, which is potentially symptomatic of a diminished accuracy in the electron-cyclotron wave power deposition and of a degraded current-drive efficiency.