Device and method for panic attack detection

Abstract

The invention discloses an apparatus and method for panic attack detection. Aiming at the problems that an existing psychological state panic attack detection mode is poor in correlation, low in precision and complex in detection equipment, firstly, three biological indexes including heart rate, skin electricity and breathing information which are strongest in correlation with panic attack are selected, and a sampling result is obtained and processed based on error correction analysis and bilateral filtering; then adopting time domain analysis to obtain a group of comprehensive feature vectors comprising heart rate, skin electricity and breathing feature information, and inputting the comprehensive feature vectors and the feature vectors which are judged to be panic attack as a data set and a verification set into a group of trained models respectively to obtain a comprehensive panic attack index; and finally, extracting the variance, extreme value and variability characteristics of the index, respectively corresponding to the psychological fluctuation state, psychological boundary and pressure-bearing state of the detected person, and outputting a three-dimensional six-quadrant coordinate distribution diagram representing the psychological state.

Description

Technical Field

[0001] The invention relates to the field of psychological stress measurement and evaluation, and in particular to a device and method for panic attack detection. Background Art

[0002] Panic attacks are usually diagnosed based on clinical evaluation and medical history, according to the patient's symptoms, frequency, duration and precipitating factors, and excluding other possible physical illnesses. Studies have shown that panic attacks are related to an imbalance of certain neurotransmitters in the brain, which is manifested in heart rate and heart rate variability, galvanic skin response, and respiratory rate and depth. A sharp increase in heart rate and a decrease in heart rate variability (HRV) are common features of panic attacks. An increase in galvanic skin GSR reflects an increase in sweat levels, which is usually related to tension and anxiety. An increase in respiratory rate and changes in breathing depth may indicate a panic attack.

[0003] Heart rate variability, as an important indicator of the function of the autonomic nervous system, reflects the balance of the heart's regulation by the sympathetic and parasympathetic nerves. In patients with panic disorder, the function of the autonomic nervous system is often disturbed, with overactivity of the sympathetic nerves and relative inhibition of the parasympathetic nerves, resulting in decreased HRV. This decreased HRV is a sign of the patient's long-term health risk because it is associated with an increased risk of cardiovascular accidents. Therefore, the measurement of HRV can provide important information for assessing the function of the autonomic nervous system and predicting health risks in patients with panic disorder.

[0004] There is also a significant association between skin electrodermal levels and panic attacks. Electrodermal activity (EDA) is a measure of the activity of the human sympathetic nervous system, which is usually manifested as physiological reactions such as sweating of the palms. During a panic attack, the sympathetic nervous system is activated due to intense emotional experiences and physiological stress responses, resulting in increased skin electrodermal levels. This increase reflects the body's immediate response to the panic attack and is part of the body's "fight or flight" mechanism. The increase in skin electrodermal levels can serve as a physiological marker of anxiety and tension during a panic attack, which can help diagnose and monitor panic disorder.

[0005] Breathing is also an important indicator of a panic attack. During a panic attack, the patient experiences intense feelings of fear and anxiety, which are often accompanied by rapid breathing or difficulty breathing, a phenomenon known as hyperventilation. Hyperventilation causes the body's carbon dioxide levels to drop, which can cause symptoms such as dizziness, palpitations, and chest tightness, which in turn exacerbate the feeling of panic. This disturbance in breathing patterns not only exacerbates the experience of a panic attack, but can also make the patient more fearful that they will not be able to control their breathing.

[0006] Using multi-dimensional wearable devices to detect panic attacks is a cutting-edge and more promising approach. Summary of the invention

[0007] The present invention provides a device and method for panic attack detection, aiming to solve the following problems: existing clinical test interviews, behavioral assessments or questionnaire measurements have high requirements on the subjects; existing measurement equipment has low accuracy; and personal factors of the tester affect the test results.

[0008] To achieve the above objectives, the present invention adopts the following technical solutions: A method for detecting panic attacks of the present invention comprises the following steps: A set of integrated sensors are used to collect the heart rate, skin electricity and breathing information of the tester; First, the data is self-checked based on error correction logic, and then the data is processed based on the bilateral filtering method to further eliminate noise and specific data; Using time domain analysis, we obtain a set of comprehensive feature vectors including heart rate, skin electricity and breathing feature information, which are input into a set of algorithm models as a data set and a validation set, respectively, to obtain a comprehensive panic attack index. Finally, the variance, extreme value and variability characteristics of the index are extracted, which correspond to the psychological fluctuation state, psychological boundary and pressure state of the tester respectively, and the output is a three-dimensional six-quadrant coordinate distribution representing the psychological state; Preferably, the data is firstly self-checked based on the error correction logic, and the data is processed by using a bilateral filtering method, including: Step 1: Normalize the heart rate, skin electricity and respiration information according to the statistical information; Step 2: Take the average of the three groups of normalized data, and filter and eliminate the data whose deviation from the average value is greater than the threshold; Step 3: Bilateral filtering is performed on the front and back numerical domains of the heart rate, skin electricity, and respiratory information based on the difference between the center point and the center point value in the neighborhood, respectively; Preferably, time domain analysis is used to obtain a set of comprehensive feature vectors including heart rate, skin electricity and breathing feature information, including: Step 1: Get the average RRavg and standard deviation of the heart rate respectively SDRR ; Mean value of skin electricity GSR avg and change rate ΔGSR; respiratory characteristic rate R esp_rate and respiratory standard deviation SD _ R esp ； Step 2: Fuse the extracted features to obtain a set of comprehensive feature vectors: F fusion =[RRavg, SDRR, GSR avg , ΔGSR, R esp_rate , S.D. _ R esp ] Preferably, the comprehensive feature vector and whether it is a panic attack are input into a set of models as a data set and a validation set, respectively, so as to obtain a comprehensive panic attack index, including: The existing statistical data, the judgment of whether panic attacks occur and the corresponding detectors RRavg, SDRR, GSR avg , ΔGSR, R esp_rate and SD _ R esp Input into the training model to get the panic attack index G panic ,get: G panic =G(RRavg, SDRR, GSR avg , ΔGSR, R esp_rate , SD _ R esp ) Preferably, extract G panic The variance, extreme value and variability characteristics correspond to the psychological fluctuation state, psychological boundary and pressure state of the tester, respectively. The output is a three-dimensional six-quadrant coordinate distribution representing the psychological state, including: Step 1: In N cycles, obtain G panic The variance, extreme value and variability characteristics of Step 2: G panic The variance, extreme value and variability characteristics of are standardized respectively; Step 3: G panic The per-unit data of the variance, extreme value and variability characteristics are output as a set of three-dimensional six-quadrant non-discrete coordinate diagrams representing the psychological state, which are used to linearly measure the subject's panic psychological fluctuation state, psychological boundaries and stress state. BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a first flow chart of an apparatus and method for panic attack detection implemented in an embodiment of the present invention; Figure 2 is a second flow chart of an apparatus and method for panic attack detection implemented by an embodiment of the present invention; Figure 3 is a third flow chart of an apparatus and method for panic attack detection implemented in an embodiment of the present invention; Figure 4This is a fourth flow chart of an apparatus and method for panic attack detection implemented according to an embodiment of the present invention.

Claims

1. A device and method for panic attack detection, characterized in that, A three-dimensional judgment index based on collecting the heart rate, skin electricity and breathing information of the tester using a set of integrated sensors is proposed.

2. A device and method for panic attack detection according to claim 1, characterized in that First, the data is self-checked based on the error correction logic, and then the data is processed based on the bilateral filtering method to further remove noise and specific data, including: Step 1: Normalize the heart rate, skin electricity and respiration information according to the statistical information; Step 2: Take the average of the three groups of normalized data, and filter and eliminate the data whose deviation from the average value is greater than the threshold; Step 3: Bilateral filtering is performed on the front and back numerical domains of the heart rate, skin electricity, and respiratory information based on the difference between the center point and the center point value in the neighborhood.

3. A device and method for panic attack detection according to claims 1 and 2, characterized in that: Using time domain analysis, we obtain a set of comprehensive feature vectors including heart rate, skin electricity and respiratory feature information, which are input into a set of models as a data set and a validation set respectively, along with whether it is a panic attack, to obtain a comprehensive panic attack index.

4. A method for detecting a mental health state according to claim 3, a device and method for detecting a panic attack, characterized in that: The variance, extreme value and variability characteristics of the index are extracted, which correspond to the psychological fluctuation state, psychological boundary and pressure state of the tester respectively, and the output is a three-dimensional six-quadrant coordinate distribution representing the psychological state, including: Step 1: In N cycles, obtain G panic The variance, extreme value and variability characteristics of Step 2: G panic The variance, extreme value and variability characteristics of are standardized respectively; Step 3: G panic The per-unit data output of the variance, extreme value and variability characteristics is a set of three-dimensional six-quadrant non-discrete coordinates representing the psychological state.

Images