Bitcoin is the cryptocurrency with the highest capitalization. It represents a striking but risky new investment due to the uncertainty generated by its high volatility, which makes it difficult to predict its behavior. This work attempts to implement a predictive model, considering scope variables and applying sentiment analysis on Tweets to predict its shortterm behavior. The approach involves data collection, analysis, treatment, and preprocessing using selection features to obtain influence variables. A model built used an LSTM (Long Short Term Memory) network whose effectiveness was measured by the appropriate metric RMSE (Root Mean Square Error). The results showed better performance for the predictive model set to one hour in the future. Finally, it is possible to predict the Bitcoin cryptocurrency behavior.