As one of the key components of electric vehicles, the Li-ion Battery Management System (BMS) is crucial to the industrialization and marketization of electric vehicles. Developing advanced and intelligent BMSs has been gathering the research interest. However, the internal states of the battery are affected by several factors, thus making the application of predictive analytics algorithms a challenging task. With the recent advances in modelling tools and diagnostics, there is an opportunity to fuse this knowledge with emerging ML techniques towards creating a battery digital twin. In this paper, we propose a data-driven digital twin of EV batteries in order to support the implementation of predictive analytics algorithms. The architecture has been modelled according to the RAMI 4.0 principles in order to provide a systematic way of modelling and development data-driven digital twins for supporting predictive analytics of battery states.