| Anxiety is a kind of extremely negative emotion,and long term anxiety is the cause of many serious physical and mental diseases.Anxiety leads to sympathetic activation and parasympathetic withdrawl,which disrupt the competitive balance of autonomic nervous activity.In daily life,real-time and accurate detection of anxiety state is conducive to timely block the physical and mental hazards of anxiety.Anxiety detection based on the autonomic neural pattern has achieved high detection precision in specific scenes.However,motion interferes with anxiety detection in arbitrary scenes of daily life,leading to the false report of anxiety.The autonomic nervous mechanism of motion interference in real-time anxiety detection has not been reported in literature,which is also the scientific problem explored in this thesis.In order to solve the above scientific problem,this thesis designed the protocol of data acquisition under typical movement states with four different intensities,analyzed the autonomic nervous activity of typical movement states,found out the typical movement states with similar autonomic patterns to the anxiety-specific autonomic patterns,determined the motion intensity by using triple-axis acceleration,recognized and eliminated the motion interference which caused false detection of anxiety state.The specific research content and results are as follows:(1)Physiological data acquisition was performed during fast running,slow running,walking and sitting,and dataset of Electrocardiogram(ECG)and triple-axis acceleration(TAA)signal under four typical motion intensities were constructed.(2)The amplitude index of TAA signals under four typical motion intensities was analyzed.It was found that there were significant differences among the amplitude index values of TAA under different motion intensities,and the amplitude index of TAA signal was effective for judging different motion intensities.(3)The autonomic nerve activities under four typical motion intensities were analyzed.Sympathetic nerve activation,parasympathetic nerve withdrawl and autonomic nervous activity imbalance were found during and right after fast and slow running,and these autonomic nervous patterns were similar to the autonomic nervous response of anxiety.Walking and sitting did not show autonomic nervous patterns similar to the autonomic nervous response of anxiety.(4)Without decreasing the true positive rate of anxiety,a multimodal system was built to automatically identify and exclude the motion interference in real-time anxiety detection.The effectiveness of the system was verified in a public speaking experiment.403 times of real anxiety state were detected,834 times of motion interference were identified,and 48 times of false anxiety detection was excluded.The effectiveness of the above system was also verified in 5 hours of real life situation,2734 times of movement interference were identified,and 895 times of false anxiety state caused by motion interference were excluded by the system.The study results showed that false anxiety detection in real life were mainly caused by the motion interference having the motion intensity similar to fast running and slow running,and autonomic patterns(i.e.,sympathetic nerve activation and parasympathetic nerve withdrawl)of these movement states similar to those of the anxiety state were the key factors causing false anxiety detection.The multi-modal real-time anxiety detection system built in this thesis identified and excluded motion interference in real life,reduced the false positive rate of real-time anxiety detection,and enhanced the application performance of real-time anxiety detection in arbitrary scenes of daily life. |
