This work has been triggered because of the need for a new way to relieve the heart. Current solutions are based on invasive systems. The main problem of such assistance is infectious risks. State of the art to define an alternative has allowed highlighting an emerging technology, i.e. the dielectric elastomer actuator (DEA). It consists of a soft silicone membrane sandwiched between two compliant electrodes and when an electric field is applied, the film is squeezed due to the electrostatic forces. The imagined solution is to place a rolled DE actuator around the aorta and through the appropriate activation, the heart is relieved but in no way replaced. Before focusing on the application and the possibility to relieve the heart, the technology is studied to highlight influencing parameters which allow increasing the performance as the energy density. An original work inspired by the thermodynamic domain to model the DEA including the phase transition is proposed. The latter is carefully explained and allows to define a figure of merit composed of the intrinsic electrical and mechanical parameters of the material. Then, the definition of several energy densities which depends on the consideration of several parameters is introduced. It is observed that the published values of energy density are too big. According to the proposed development, a more realistic one is about ten times lower. Once the definitions and the model, allowing to study a planar dielectric elastomer actuator are provided, the specific energy density of such actuator is studied in several conditions. An external biasing element such as a constant load and an initial pre-stretch of the membrane are analysed. The results show that tracking the maximal strains does not necessary means that the optimal maximal energy density is reached. In the literature, it has been demonstrated that the negative biasing element allows improving the displacement of DEAs. Through the proposed definitions of energy density, this biasing element is studied. The obtained stretches and the specific energy densities are two times bigger than with the constant load. Due to the application previously introduced, a cylindrical spring is proposed which owns this negative characteristic. After the study of the influence of the geometrical parameters on its force-displacement characteristic, a prototype is proposed. The latter is coupled to a DEA for which the global system is analytically modelled. The prototype is validated with the measurements and FE analysis. An important observation concerns the special deformation of the spring. Through an analogy with the stability of the elastomer, an explication is provided concerning the global deformation. Finally, to determine the feasibility to use the DEA technology as cardiac assistance, a tubular assist device based DEA is analysed. Through a lumped parameters model which allows simulating the cardiac cycle, different configurations of activation are studied. It has been decided to analyse the possibility to use such technology to relieve the left ventricle and not to propose an optimised solution. Thus, the previously designed spring is not considered. The results show that two main configurations could be used to unload the left ventricle. The first one decreases the energy provided by the heart and the other increases the cardiac output which seems an attractive solution.