Recent advancements in lithium batteries technology are a major catalyst for the development of smart energy grids. Despite their large use and improvements in multiple fields, there are still some concerns regarding safety issues and fast degradation processes. For these reasons, the development of accurate State-Of-Health (SOH) estimation and prediction algorithm is essential to guarantee reliability, availability and safety of both the battery pack and the entire smart grid. Usually, most of the papers in literature deal with this problem using a single feature, which is the discharge capacity of the battery. However, considering the entire Prognostic and Health Management (PHM) framework, there are also other battery parameters that need to be estimated and predicted to ensure safe and reliable operations. Trying to fill this gap, this paper presents the comparison between three different LSTM-based (Long Short-Term Memory) algorithms for simultaneous and concatenated estimation and prediction of multiple battery features, including charge and discharge capacity, internal resistance, charge time and cell's temperature. The validation of the procedure is carried out on a large publicly available battery degradation dataset pointing out significant performances in the prediction of all features.