The assessment of brain dynamic status necessitates the knowledge of neural indicators or biomarkers which should be preferably noninvasive, simple to record and analyze, simultaneously robust and sensitive to changes in brain function in ecological situations. Nevertheless, until now, most studies aiming to identify brain biomarkers have mainly focused on biochemical, cellular, and genetic aspects related to brain disorders. Moreover, most of the findings were mainly based on a population analysis without focusing on the decoding of brain biomarkers at the individual level. Therefore, here, we investigated if it was possible to decode a biomarker able to predict the individual level of performance during the learning of a new tool. The magnetoencephalography (MEG) signals and hand kinematics during the acquisition of a novel screen cursor-hand relationship were measured for five subjects. The alpha power (9-13 Hz) has been computed by integrating the output of a band-pass fourth-order Butterworth filter. A moving average method has been applied to both MEG and kinematics signals. The findings revealed that, at the individual level¸ the changes in alpha power (9-13 Hz) computed through learning at the frontal, temporal and parietal regions during movement preparation correlated correctly the evolution of the movement kinematics level in the subsequent performance. These results suggest that the presented approach can be used to decode on an individual basis, the internal cognitive-sensorimotor state of the brain during the learning/adaptation to a new tool without having access to the actual performance. This approach may be useful for brain computer interface and diagnostic of individual having specific motor disabilities.