Attention is a psychological and neurophysiological phenomenon, and associated with task difficulty. With the development of electroencephalography (EEG) detection tools, mature and affordable mobile brainwave sensors have become available. Therefore, this study aims to predict attentional status with the difficulty of work based on the EEG signals, and apply the process of prediction to an Android application. In the study, 10 participants performed different tasks with NeuroSky's MindWave EEG device, which can collect participants' EEG signals. K-means clustering algorithm was used to estimate the boundary of different attentional status. Moreover, Q-value was used to evaluate the question difficulty. 20 participants were recorded gender, the EEG signals, question difficulty, results, and answer time, which were participant predictors. The attentional status was predicted by support vector machine (SVM) classifier. The findings indicated that early identity of attentional status is essential to the learning efficiency, and provided a guide to the design of task.