Wi-Fi and Bluetooth are two wireless technologies, available in every smart-phone, tablet, and laptop. Wi-Fi Access Points (APs) and Bluetooth beacons are deployed in most indoor environments to provide service for the Internet of Things (IoT) applications. Although, Bluetooth and Wi-Fi target different applications, they both share the 2.4 GHz frequency band. The re-transmissions caused by interference with Wi-Fi packets is costly for BLE in terms of energy consumption. Techniques such as Adaptive Frequency Hopping (AFH) in BLE addresses this problem. However, the static nature of AFH is not performing well for highly dynamic environments. Therefore, there is a need for a predictive model to optimize the spectrum usage. In this paper, we propose a machine learning model based on Long Short-Term Memory (LSTM) to predict the wireless activities in the 2.4 GHz frequency band and its impact on BLE channels. We apply the proposed model to analyze the Wi-Fi interference trend on these channels. The Root Mean Squared Error (RMSE) results for several experiments on both channels indicate the high performance of the proposed LSTM model over Auto Regressive Integrated Moving Average (ARIMA) model. This improvement is significant up to approximately 50% reduction in error.