Implementation of Distracted Estimation System based on Sensor Fusion through Correlation Analysis with Concentration

Sitting for an extended time may cause a serious chronic disease such as a musculoskeletal disorder, or a cardiovascular disease, diabetes, or obesity. Because a consistently improper posture from early childhood to adolescence can have a number of undesirable effects on the formation of the musculoskeletal structure, learning to maintain a correct posture should be emphasized. A consistently improper posture can not only cause physical problems, it may also lead to emotional issues such as distractions, an attention deficit, and hyperactivity, and the possibility of a low efficiency and performance on assignments is high when the students have a low concentration. The present study implemented a distracted estimation system based on sensor fusion through correlation analysis with concentration that could estimate the level of distraction and prevent musculoskeletal diseases caused by poor sitting posture habits in daily life. The implemented system was designed in the form of a sitting cushion to reflect the ethological movements and characteristics of a sitting position that modern people spend a large amount of time in, and can be easily applied to existing chairs. Both algorithms installed in the system detected the center of gravity of the seated person and displayed positional changes that occurred based on the intensity of the postural changes when moving; thus, simultaneous determination of posture and impulsive behavior was possible. To evaluate the system performance, a posture determination evaluation was conducted, along with distraction estimation according to the rate of changes in posture that occur in everyday life. In addition, to evaluate performance in daily life, a movie-watching scenario was set up, and the distracted-limit estimation and concentration indices according to the rate of changes in posture were comparatively evaluated by reviewing a video of the subjects. The results of the posture determination performance evaluation through 100 posture repetitions on 10 subjects showed a high detection performance of 99.04%. The Pearson’s correlation coefficient results showed a high correlation coefficient (inverse) of r = −0.975076 and a P-VALUE = [Formula: see text]. This experiment objectively confirmed the correlation between the DLE Index (based on postural change) and the CI Index (based on EEG).